Handbook of Solvents

HANDBOOK OF SOLVENTS George Wypych, Editor ChemTec Publishing Toronto − New York 2001 Published by ChemTec Publishi...

Author: George Wypych

183 downloads 1930 Views 19MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

HANDBOOK OF

SOLVENTS George Wypych, Editor

ChemTec Publishing

Toronto − New York 2001

Published by ChemTec Publishing 38 Earswick Drive, Toronto, Ontario M1E 1C6, Canada Co-published by William Andrew Inc. 13 Eaton Avenue, Norwich, N Y 13815, USA © Chem Tec Publishing, 2001 ISBN 1-895198-24-0 All rights reserved. No part of this publication may be reproduced, stored or transmitted in any form or by any means without written permission of copyright owner. No responsibility is assumed by the Author and the Publisher for any injury or/and damage to persons or properties as a matter of products liability, negligence, use, or operation of any methods, product ideas, or instructions published or suggested in this book.

Canadian Cataloguing in Publication Data Handbook of Solvents Includes bibliographical references and index ISBN 1-895198-24-0 (ChemTec Publishing) ISBN 0-8155-1458-1 (William Andrew Inc.) Library of Congress Catalog Card Number: 00-106798 1. Solvents--Handbooks, manuals, etc. I. Wypych, George TP247.5.H35 2000

661’.807

C00-900997-3

Printed in Canada by Transcontinental Printing Inc., 505 Consumers Rd. Toronto, Ontario M2J 4V8

Table of Contents Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvii GEORGE WYPYCH

1

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 2.1

FUNDAMENTAL PRINCIPLES GOVERNING SOLVENTS USE . . . . 7 Solvent effects on chemical systems . . . . . . . . . . . . . . . . . . . . . . . 7

2.1.1 2.1.2 2.1.2.1 2.1.2.2 2.1.2.3 2.1.2.4 2.1.2.5 2.1.2.6 2.1.3 2.1.3.1 2.1.3.2 2.1.3.3 2.1.3.4 2.1.3.5 2.1.4 2.1.4.1 2.1.4.2 2.1.4.3 2.1.5 2.2

Historical outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Classification of solute-solvent interactions . . . . . . . . . . . . . . . . . . Electrostatic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Repulsion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Specific interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrophobic interactions. . . . . . . . . . . . . . . . . . . . . . . . . . . . Modelling of solvent effects . . . . . . . . . . . . . . . . . . . . . . . . . . Computer simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Continuum models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cavity surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Supermolecule models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application example: glycine in solution . . . . . . . . . . . . . . . . . . . Thermodynamic and kinetic characteristics of chemical reactions in solution Solvent effects on chemical equilibria . . . . . . . . . . . . . . . . . . . . . Solvent effects on the rate of chemical reactions. . . . . . . . . . . . . . . . Example of application: addition of azide anion to tetrafuranosides. . . . . . Solvent catalytic effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular design of solvents . . . . . . . . . . . . . . . . . . . . . . . . . .

2.2.1 2.2.2 2.2.3 2.2.4 2.2.5 2.3

Molecular design and molecular ensemble design . . . . . . . . . . . . . . . From prediction to design . . . . . . . . . . . . . . . . . . . . . . . . . . . Improvement in prediction method. . . . . . . . . . . . . . . . . . . . . . . Role of molecular simulation. . . . . . . . . . . . . . . . . . . . . . . . . . Model system and paradigm for design . . . . . . . . . . . . . . . . . . . . Appendix. Predictive equation for the diffusion coefficient in dilute solution Basic physical and chemical properties of solvents . . . . . . . . . . . . . .

2.3.1 2.3.2 2.3.3 2.3.4 2.3.5 2.3.6 2.3.7 2.3.8 2.3.9 2.3.10 2.3.11 2.3.12 2.3.13 2.3.14 2.3.15

Molecular weight and molar volume . . . . . . . Boiling and freezing points. . . . . . . . . . . . Specific gravity . . . . . . . . . . . . . . . . . . Refractive index . . . . . . . . . . . . . . . . . Vapor density and pressure. . . . . . . . . . . . Solvent volatility . . . . . . . . . . . . . . . . . Flash point . . . . . . . . . . . . . . . . . . . . Flammability limits. . . . . . . . . . . . . . . . Sources of ignition and autoignition temperature Heat of combustion (calorific value) . . . . . . . Heat of fusion. . . . . . . . . . . . . . . . . . . Electric conductivity . . . . . . . . . . . . . . . Dielectric constant (relative permittivity) . . . . Occupational exposure indicators . . . . . . . . Odor threshold . . . . . . . . . . . . . . . . . .

CHRISTIAN REICHARDT

ESTANISLAO SILLA, ARTURO ARNAU, IÑAKI TUÑÓN

. . . . . . . . . . . . . . . . . . . .

7 10 11 12 13 14 15 16 17 18 20 21 22 23 27 27 28 30 32 36

. . . . . . .

36 37 38 39 40 41 42

. . . . . . . . . . . . . . .

43 44 46 47 48 49 50 51 52 54 54 54 54 56 56

KOICHIRO NAKANISHI

GEORGE WYPYCH

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

ii

2.3.16 2.3.17 2.3.18 2.3.19 2.3.20 3

Handbook of Solvents

3.1

Toxicity indicators . . . . . . . . . . . . . . . . Ozone-depletion and creation potential . . . . . Oxygen demand . . . . . . . . . . . . . . . . . Solubility . . . . . . . . . . . . . . . . . . . . . Other typical solvent properties and indicators . PRODUCTION METHODS, PROPERTIES, AND MAIN APPLICATIONS . . . . . . . . . Definitions and solvent classification . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

57 58 58 58 60

3.2

Overview of methods of solvent manufacture . . . . . . . . . . . . . . . . . . 69

3.3

Solvent properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

3.3.1 3.3.1.1 3.3.1.2 3.3.2 3.3.3 3.3.4 3.3.5 3.3.6 3.3.7 3.3.8 3.3.9 3.3.10 3.3.11 3.3.11 3.3.12 3.3.13 3.3.14 3.4

Hydrocarbons. . . . . . . . . . . . . . . . . . . . Aliphatic hydrocarbons. . . . . . . . . . . . . . . Aromatic hydrocarbons. . . . . . . . . . . . . . . Halogenated hydrocarbons . . . . . . . . . . . . . Nitrogen-containing compounds (nitrates, nitriles) Organic sulfur compounds . . . . . . . . . . . . . Monohydric alcohols . . . . . . . . . . . . . . . . Polyhydric alcohols. . . . . . . . . . . . . . . . . Phenols . . . . . . . . . . . . . . . . . . . . . . . Aldehydes . . . . . . . . . . . . . . . . . . . . . Ethers . . . . . . . . . . . . . . . . . . . . . . . . Glycol ethers . . . . . . . . . . . . . . . . . . . . Ketones . . . . . . . . . . . . . . . . . . . . . . . Acids . . . . . . . . . . . . . . . . . . . . . . . . Amines . . . . . . . . . . . . . . . . . . . . . . . Esters . . . . . . . . . . . . . . . . . . . . . . . . Comparative analysis of all solvents . . . . . . . . Terpenes . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

75 75 76 78 79 80 81 83 84 85 86 87 88 90 91 92 94 96

3.4.1 3.4.2 3.4.3 3.4.4 3.4.5 4

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

96 96 96 97 97

4.1

Definitions and nomenclature . . . . . . . . . . . . . . . . . . Occurrence . . . . . . . . . . . . . . . . . . . . . . . . . . . . General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Toxicology . . . . . . . . . . . . . . . . . . . . . . . . . . . . Threshold limit values . . . . . . . . . . . . . . . . . . . . . . GENERAL PRINCIPLES GOVERNING DISSOLUTION OF MATERIALS IN SOLVENTS . . . . . . . . . . . . . . . Simple solvent characteristics . . . . . . . . . . . . . . . . . .

4.1.1 4.1.2 4.1.3 4.1.4 4.1.5 4.1.6 4.1.7 4.2

Solvent power . . . . . . . . . . . . . . . . . . One-dimensional solubility parameter approach . Multi-dimensional approaches . . . . . . . . . . Hansen’s solubility . . . . . . . . . . . . . . . . Three-dimensional dualistic model. . . . . . . . Solubility criterion . . . . . . . . . . . . . . . . Solvent system design . . . . . . . . . . . . . . Effect of system variables on solubility . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

101 103 110 112 116 119 120 124

4.2.1 4.2.2 4.2.3 4.2.4 4.2.5 4.2.6

General considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . Chemical structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Flexibility of a polymer chain . . . . . . . . . . . . . . . . . . . . . . . Crosslinking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Temperature and pressure . . . . . . . . . . . . . . . . . . . . . . . . . Methods of calculation of solubility based on thermodynamic principles .

. . . . . .

. . . . . .

. . . . . .

124 126 127 128 128 130

. . . . . . . . . . . . . . . . 65 . . . . . . . . . . . . . . . . 65

GEORGE WYPYCH GEORGE WYPYCH GEORGE WYPYCH

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

TILMAN HAHN, KONRAD BOTZENHART, FRITZ SCHWEINSBERG

. . . . . . . . 101 . . . . . . . . 101

VALERY YU. SENICHEV, VASILIY V. TERESHATOV

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .


Table of contents

iii

4.3

Polar solvation dynamics: Theory and simulations . . . . . . . . . . . . . . . 132

4.3.1 4.3.2 4.3.3 4.3.4

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Continuum dielectric theory of solvation dynamics . . . . . . . . . . . Linear response theory of solvation dynamics . . . . . . . . . . . . . . Numerical simulations of solvation in simple polar solvents: The simulation model . . . . . . . . . . . . . . . . . . . . . . . . . . Numerical simulations of solvation in simple polar solvents: Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . Solvation in complex solvents . . . . . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods for the measurement of solvent activity of polymer solutions .

ABRAHAM NITZAN

4.3.5 4.3.6 4.3.7 4.4

. . . . 132 . . . . 133 . . . . 136 . . . . 138 . . . .

. . . .

. . . .

. . . .

140 144 145 146

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

146 149 154 154 155 170 175 178 178 181 184 185 186 188 189 193

CHRISTIAN WOHLFARTH

4.4.1 4.4.2 4.4.3 4.4.3.1 4.4.3.1.1 4.4.3.1.2 4.4.3.1.3 4.4.3.2 4.4.3.2.1 4.4.3.2.2 4.4.3.2.3 4.4.3.2.4 4.4.3.2.5 4.4.3.2.6 4.4.3.2.7 4.4.3.2.8 4.4.4 4.4.4.1 4.4.4.2 4.4.4.3 5 5.1

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Necessary thermodynamic equations. . . . . . . . . . . . . . . . . Experimental methods, equipment and data reduction . . . . . . . . Vapor-liquid equilibrium (VLE) measurements . . . . . . . . . . . Experimental equipment and procedures for VLE-measurements . . Primary data reduction . . . . . . . . . . . . . . . . . . . . . . . . Comparison of experimental VLE-methods . . . . . . . . . . . . . Other measurement methods . . . . . . . . . . . . . . . . . . . . . Membrane osmometry . . . . . . . . . . . . . . . . . . . . . . . . Light scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . X-ray scattering. . . . . . . . . . . . . . . . . . . . . . . . . . . . Neutron scattering . . . . . . . . . . . . . . . . . . . . . . . . . . Ultracentrifuge . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cryoscopy (freezing point depression of the solvent) . . . . . . . . Liquid-liquid equilibrium (LLE) . . . . . . . . . . . . . . . . . . . Swelling equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . Thermodynamic models for the calculation of solvent activities of polymer solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . Models for residual chemical potential and activity coefficient in the liquid phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fugacity coefficients from equations of state . . . . . . . . . . . . Comparison and conclusions . . . . . . . . . . . . . . . . . . . . . Appendix 4.4A . . . . . . . . . . . . . . . . . . . . . . . . . . . . SOLUBILITY OF SELECTED SYSTEMS AND INFLUENCE OF SOLUTES . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental methods of evaluation and calculation of solubility parameters of polymers and solvents. Solubility parameters data . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . 195 . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

196 207 214 223

. . . . . . 243 . . . . . . 243


5.1.1 5.1.1.1 5.1.1.2 5.1.1.3 5.1.1.4 5.1.2 5.2

Experimental evaluation of solubility parameters of liquids . . . . Direct methods of evaluation of the evaporation enthalpy . . . . Indirect methods of evaluation of evaporation enthalpy . . . . . . Static and quasi-static methods of evaluation of pair pressure . . . Kinetic methods . . . . . . . . . . . . . . . . . . . . . . . . . . Methods of experimental evaluation and calculation of solubility parameters of polymers. . . . . . . . . . . . . . . . . . . . . . . Prediction of solubility parameter . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

5.2.1 5.2.2 5.2.2.1 5.2.2.2 5.2.3 5.3

Solubility parameter of polymers . . . . . . . . . . . . . . . . . . . . . Glass transition in polymers . . . . . . . . . . . . . . . . . . . . . . . . Glass transition enthalpy . . . . . . . . . . . . . . . . . . . . . . . . . . Cp jump at the glass transition . . . . . . . . . . . . . . . . . . . . . . . Prediction from thermal transition enthalpies . . . . . . . . . . . . . . . Methods of calculation of solubility parameters of solvents and polymers

. . . . .

. . . . .

. . . . .

243 243 244 245 245

. . . . . . . 246 . . . . . . . 253

NOBUYUKI TANAKA


. . . . . .

. . . . . .

. . . . . .

253 254 254 256 258 261

iv


5.4

Mixed solvents, a way to change the polymer solubility. . . . . . . . . . . . . 267

5.4.1 5.4.2 5.4.3 5.4.4

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solubility-cosolvency phenomenon . . . . . . . . . . . . . . . . . . . . . . . New cosolvents effects. Solubility behavior . . . . . . . . . . . . . . . . . . . Thermodynamical description of ternary systems. Association equilibria theory of preferential adsorption . . . . . . . . . . . . . . . . . . . . . . . . . Polymer structure of the polymer dependence of preferential adsorption. Polymer molecular weight and tacticity dependence of preferential adsorption. The phenomenological theory of solvent effects in mixed solvent systems . . .

LIGIA GARGALLO AND DEODATO RADIC

5.4.5 5.5

267 268 273 274 277 281

KENNETH A. CONNORS

5.5.1 5.5.2 5.5.2.1 5.5.2.2 5.5.2.3 5.5.2.4 5.5.2.5 5.5.3 5.5.3.1 5.5.3.2 5.5.3.3 5.5.3.4 5.5.3.5 5.5.3.6 5.5.4 5.5.4.1 5.5.4.2 5.5.4.3 5.5.4.4

6 6.1

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The intersolute effect: solute-solute interactions . . . . . . . . . . . . . . . The solvation effect: solute-solvent interaction . . . . . . . . . . . . . . . The general medium effect: solvent-solvent interactions . . . . . . . . . . The total solvent effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solubility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Surface tension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electronic absorption spectra. . . . . . . . . . . . . . . . . . . . . . . . . Complex formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chemical kinetics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Liquid chromatography. . . . . . . . . . . . . . . . . . . . . . . . . . . . Interpretations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ambiguities and anomalies. . . . . . . . . . . . . . . . . . . . . . . . . . A modified derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . Interpretation of parameter estimates. . . . . . . . . . . . . . . . . . . . . Confounding effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solute-solute interactions. . . . . . . . . . . . . . . . . . . . . . . . . . . Coupling of general medium and solvation effects . . . . . . . . . . . . . The cavity surface area . . . . . . . . . . . . . . . . . . . . . . . . . . . . The role of interfacial tension . . . . . . . . . . . . . . . . . . . . . . . . SWELLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Modern views on kinetics of swelling of crosslinked elastomers in solvents

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

281 281 281 282 283 284 285 285 285 288 290 291 295 298 298 298 299 300 301 301 301 301 302 305 305

6.1.1 6.1.2 6.1.3 6.1.4 6.1.5 6.2

Introduction. . . . . . . . . . . . . . . . . . . . . . Formulation of swelling for a plane elastomer layer . Diffusion kinetics of plane layer swelling . . . . . . Experimental study of elastomer swelling kinetics . Conclusions. . . . . . . . . . . . . . . . . . . . . . Equilibrium swelling in binary solvents . . . . . . .

. . . . . .

. . . . . .

305 306 310 314 317 318

6.3

Swelling data on crosslinked polymers in solvents . . . . . . . . . . . . . . . 327

6.4

Influence of structure on equilibrium swelling. . . . . . . . . . . . . . . . . . 331

7 7.1

SOLVENT TRANSPORT PHENOMENA . . . . . . . . . . . . . . . . . . 339 Introduction to diffusion, swelling, and drying . . . . . . . . . . . . . . . . . 339

7.1.1 7.1.2 7.1.3 7.2

Diffusion . . . . . . . . . . . . . . . . . . . . . . . . Swelling . . . . . . . . . . . . . . . . . . . . . . . . Drying . . . . . . . . . . . . . . . . . . . . . . . . . Bubbles dynamics and boiling of polymeric solutions .

7.2.1

Rheology of polymeric solutions and bubble dynamics . . . . . . . . . . . . . 356

E. YA. DENISYUK, V. V. TERESHATOV

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

VASILIY V. TERESHATOV, VALERY YU. SENICHEV VASILIY V. TERESHATOV, VALERY YU. SENICHEV VASILIY V. TERESHATOV, VALERY YU. SENICHEV

GEORGE WYPYCH

SEMYON LEVITSKY, ZINOVIY SHULMAN

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

339 344 348 356

Table of contents

v

7.2.1.1 7.2.1.2 7.2.2 7.2.3 7.3

Rheological characterization of solutions of polymers. . . . . . . Dynamic interaction of bubbles with polymeric liquid . . . . . . Thermal growth of bubbles in superheated solutions of polymers Boiling of macromolecular liquids . . . . . . . . . . . . . . . . . Drying of coated film. . . . . . . . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

356 363 372 377 386

7.3.1 7.3.2 7.3.2.1 7.3.2.2 7.3.2.3 7.3.2.4 7.3.2.5

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theory for the drying . . . . . . . . . . . . . . . . . . . . . . . . . . . Simultaneous heat and mass transfer . . . . . . . . . . . . . . . . . . . Liquid-vapor equilibrium. . . . . . . . . . . . . . . . . . . . . . . . . Heat and mass transfer coefficient . . . . . . . . . . . . . . . . . . . . Prediction of drying rate of coating . . . . . . . . . . . . . . . . . . . Drying regimes: constant drying rate period (CDRP) and falling drying rate period (FDRP) . . . . . . . . . . . . . . . . . . . . . . . . Measurement of the drying rate of coated film. . . . . . . . . . . . . . Thermo-gravimetric analysis . . . . . . . . . . . . . . . . . . . . . . . Rapid scanning FT-IR spectrometer analysis . . . . . . . . . . . . . . High-airflow drying experiment using flame ionization detector (FID) total hydrocarbon analyzer . . . . . . . . . . . . . . . . . . . . . . . . Measurement of drying rate in the production scale dryer . . . . . . . . Miscellaneous . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Drying of coated film with phase separation . . . . . . . . . . . . . . . Drying defects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Internal stress induced defects . . . . . . . . . . . . . . . . . . . . . . Surface tension driven defects . . . . . . . . . . . . . . . . . . . . . . Defects caused by air motion and others . . . . . . . . . . . . . . . . . Control of lower explosive level (LEL) in a multiple zone dryer . . . . INTERACTIONS IN SOLVENTS AND SOLUTIONS . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

386 388 388 389 390 392

. . . .

. . . .

. . . .

. . . .

394 396 396 399

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

401 404 407 407 409 409 412 414 414 419

Solvents and solutions as assemblies of interacting molecules . . . . . . Basic simplifications of the quantum model . . . . . . . . . . . . . . . . Cluster expansion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Two-body interaction energy: the dimer . . . . . . . . . . . . . . . . . . Decomposition of the interaction energy of a dimer: variational approach The electrostatic term. . . . . . . . . . . . . . . . . . . . . . . . . . . . The induction term . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The exchange term . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The charge transfer term . . . . . . . . . . . . . . . . . . . . . . . . . . The dispersion term . . . . . . . . . . . . . . . . . . . . . . . . . . . . The decomposition of the interaction energy through a variational approach: a summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basis set superposition error and counterpoise corrections . . . . . . . . Perturbation theory approach. . . . . . . . . . . . . . . . . . . . . . . . Modeling of the separate components of ∆E . . . . . . . . . . . . . . . . The electrostatic term. . . . . . . . . . . . . . . . . . . . . . . . . . . . The induction term . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The dispersion term . . . . . . . . . . . . . . . . . . . . . . . . . . . . The exchange (or repulsion) term . . . . . . . . . . . . . . . . . . . . . The other terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A conclusive view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The relaxation of the rigid monomer constraint . . . . . . . . . . . . . . Three- and many-body interactions . . . . . . . . . . . . . . . . . . . . Screening many-body effects. . . . . . . . . . . . . . . . . . . . . . . . Effective interaction potentials . . . . . . . . . . . . . . . . . . . . . . . The variety of interaction potentials . . . . . . . . . . . . . . . . . . . . Theoretical and computing modeling of pure liquids and solutions . . . . Physical models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

419 420 424 424 426 426 428 428 429 430

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

432 433 436 441 441 445 446 447 448 448 449 451 453 454 456 461 461

SEUNG SU KIM AND JAE CHUN HYUN

7.3.3 7.3.3.1 7.3.3.2 7.3.3.3 7.3.3.4 7.3.4 7.3.4.1 7.3.4.2 7.3.4.2.1 7.3.4.2.2 7.3.4.2.3 7.3.4.3 8

JACOPO TOMASI, BENEDETTA MENNUCCI, CHIARA CAPPELLI

8.1 8.2 8.3 8.4 8.4.1

8.4.2 8.4.3 8.4.4

8.4.5 8.5 8.6 8.7 8.7.1

vi

8.7.1.1 8.7.1.2 8.7.2 8.7.2.1 8.7.2.2

8.7.3 8.7.3.1 8.8

8.9 8.9.1 8.9.2 8.9.3 9


Integral equation methods . . . . . . . . . . . . Perturbation theories . . . . . . . . . . . . . . . Computer simulations . . . . . . . . . . . . . . Car-Parrinello direct QM simulation . . . . . . . Semi-classical simulations . . . . . . . . . . . . Molecular dynamics . . . . . . . . . . . . . . . Monte Carlo . . . . . . . . . . . . . . . . . . . QM/MM . . . . . . . . . . . . . . . . . . . . . Continuum models . . . . . . . . . . . . . . . . QM-BE methods: the effective Hamiltonian . . . Practical applications of modeling . . . . . . . . Dielectric constant . . . . . . . . . . . . . . . . Thermodynamical properties . . . . . . . . . . . Compressibilities . . . . . . . . . . . . . . . . . Relaxation times and diffusion coefficients . . . Shear viscosity . . . . . . . . . . . . . . . . . . Liquid surfaces . . . . . . . . . . . . . . . . . . The basic types of liquid surfaces . . . . . . . . Systems with a large surface/bulk ratio . . . . . Studies on interfaces using interaction potentials MIXED SOLVENTS . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

465 467 468 470 472 472 473 478 479 482 487 487 490 490 491 492 492 493 495 497 505

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chemical interaction between components in mixed solvents . . . . . . Processes of homomolecular association. . . . . . . . . . . . . . . . . Conformic and tautomeric equilibrium. Reactions of isomerization. . . Heteromolecular association . . . . . . . . . . . . . . . . . . . . . . . Heteromolecular associate ionization . . . . . . . . . . . . . . . . . . Electrolytic dissociation (ionic association) . . . . . . . . . . . . . . . Reactions of composition. . . . . . . . . . . . . . . . . . . . . . . . . Exchange interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . Amphoterism of mixed solvent components . . . . . . . . . . . . . . . Amphoterism of hydrogen acids . . . . . . . . . . . . . . . . . . . . . Amphoterism of L-acids . . . . . . . . . . . . . . . . . . . . . . . . . Amphoterism in systems H-acid-L-acid . . . . . . . . . . . . . . . . . Amphoterism in binary solutions amine-amine . . . . . . . . . . . . . Physical properties of mixed solvents . . . . . . . . . . . . . . . . . . The methods of expression of mixed solvent compositions . . . . . . . Permittivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Density, molar volume . . . . . . . . . . . . . . . . . . . . . . . . . . Electrical conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . Physical characteristics of the mixed solvents with chemical interaction between components . . . . . . . . . . . . . . . . . . . . . . . . . . . Permittivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Density, molar volume . . . . . . . . . . . . . . . . . . . . . . . . . . Conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chemical properties of mixed solvents. . . . . . . . . . . . . . . . . . Autoprotolysis constants . . . . . . . . . . . . . . . . . . . . . . . . . Solvating ability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Donor-acceptor properties . . . . . . . . . . . . . . . . . . . . . . . . Mixed solvent influence on the chemical equilibrium . . . . . . . . . . General considerations . . . . . . . . . . . . . . . . . . . . . . . . . . Mixed solvent effect on the position of equilibrium of homomolecular association process . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mixed solvent influence on the conformer equilibrium . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

505 505 505 506 507 507 508 508 509 509 509 509 510 510 511 511 513 515 516 516

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

517 518 519 521 522 524 524 526 527 527 527

Y. Y. FIALKOV, V. L. CHUMAK

9.1 9.2 9.2.1 9.2.2 9.2.3 9.2.4 9.2.5 9.2.6 9.2.7 9.2.8 9.2.8.1 9.2.8.2 9.2.8.3 9.2.8.4 9.3 9.3.1 9.3.1.1 9.3.1.2 9.3.1.3 9.3.1.4 9.3.2 9.3.2.1 9.3.2.2 9.3.2.3 9.3.2.4 9.3.3 9.3.3.1 9.3.3.2 9.3.3.3 9.4 9.4.1 9.4.2 9.4.3

. . . . 529 . . . . 530

Table of contents

9.4.4 9.4.4.1 9.4.5 9.4.6

9.4.7 9.5 10 10.1

Solvent effect on the process of heteromolecular association . . . . . . Selective solvation. Resolvation . . . . . . . . . . . . . . . . . . . . . Mixed solvent effect on the ion association process . . . . . . . . . . . Solvent effect on exchange interaction processes . . . . . . . . . . . . Systems with non-associated reagents . . . . . . . . . . . . . . . . . . Systems with one associated participant of equilibrium . . . . . . . . . Systems with two associated participants of equilibrium . . . . . . . . Mixed solvent effect on processes of complex formation . . . . . . . . The mixed solvent effect on the chemical equilibrium thermodynamics ACID-BASE INTERACTIONS . . . . . . . . . . . . . . . . . . . . General concept of acid-base interactions . . . . . . . . . . . . . . . .

vii

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

532 538 546 552 552 553 553 556 557 565 565

GEORGE WYPYCH

10.2

Effect of polymer/solvent acid-base interactions: relevance to the aggregation of PMMA . . . . . . . . . . . . . . . . . . . . . . . . . . . . 570

10.2.1 10.2.1.1 10.2.1.1.1 10.2.1.1.2 10.2.1.1.3 10.2.1.2 10.2.1.2.1 10.2.1.2.2 10.2.1.2.3 10.2.1.2.4 10.2.1.2.5 10.2.1.2.6 10.2.2 10.2.2.1 10.2.2.2

Recent concepts in acid-base interactions . . . . . . . . . . . . . . . . . . . . The nature of acid-base molecular interactions . . . . . . . . . . . . . . . . . The original Lewis definitions . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular Orbital (MO) approach to acid-base reactions . . . . . . . . . . . . The case of hydrogen bonding . . . . . . . . . . . . . . . . . . . . . . . . . . Quantitative determination of acid-base interaction strength . . . . . . . . . . Perturbation theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hard-Soft Acid-Base (HSAB) principle . . . . . . . . . . . . . . . . . . . . . Density functional theory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effect of ionocity and covalency: Drago’s concept . . . . . . . . . . . . . . . Effect of amphotericity of acid-base interaction: Gutmann’s numbers . . . . . Spectroscopic measurements: Fowkes’ approach . . . . . . . . . . . . . . . . Effect of polymer/solvent interactions on aggregation of stereoregular PMMA Aggregation of stereoregular PMMA . . . . . . . . . . . . . . . . . . . . . . Relation between the complexing power of solvents and their acid-base properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Influence of the nature of the solvent on the α and β-relaxations of conventional PMMA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dielectric spectroscopy results . . . . . . . . . . . . . . . . . . . . . . . . . . Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent effects based on pure solvent scales . . . . . . . . . . . . . . . . . . .

S. BISTAC, M. BROGLY

10.2.3 10.2.3.1 10.2.3.2 10.2.4 10.3

570 571 571 571 573 574 574 574 575 576 577 578 578 578 579 581 581 581 582 583

JAVIER CATALÁN

10.3.1 10.3.2 10.3.3 10.3.3.1 10.3.3.2 10.3.3.3 10.3.3.4 10.3.3.5 10.3.3.6 10.3.3.7 10.3.3.8 10.3.4 10.3.5 10.3.6 10.3.7 10.3.7.1 10.3.7.2

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The solvent effect and its dissection into general and specific contributions . Characterization of a molecular environment with the aid of the probe/homomorph model. . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-parameter solvent scales: the Y, G, ET(30), Py , Z, χR, Φ, and S' scales. The solvent ionizing power scale or Y scale . . . . . . . . . . . . . . . . . . The G values of Allerhand and Schleyer . . . . . . . . . . . . . . . . . . . . The ET(30) scale of Dimroth and Reichardt . . . . . . . . . . . . . . . . . . The Py scale of Dong and Winnick . . . . . . . . . . . . . . . . . . . . . . . The Z scale of Kosower . . . . . . . . . . . . . . . . . . . . . . . . . . . . The χR scale of Brooker . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Φ scale of Dubois and Bienvenüe . . . . . . . . . . . . . . . . . . . . . The S' scale of Drago . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent polarity: the SPP scale . . . . . . . . . . . . . . . . . . . . . . . . . Solvent basicity: the SB scale . . . . . . . . . . . . . . . . . . . . . . . . . Solvent acidity: the SA scale . . . . . . . . . . . . . . . . . . . . . . . . . . Applications of the pure SPP, SA and SB scales. . . . . . . . . . . . . . . . Other reported solvents scales . . . . . . . . . . . . . . . . . . . . . . . . . Treatment of the solvent effect in: . . . . . . . . . . . . . . . . . . . . . . .

. 583 . 584 . . . . . . . . . . . . . . . .

585 587 587 588 588 589 589 590 590 591 591 600 601 605 605 608

viii


10.3.7.2.1 10.3.7.2.2 10.3.7.2.3 10.3.7.2.4 10.3.7.3 10.4

Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electrochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mixtures of solvents. Understanding the preferential solvation model Acid-base equilibria in ionic solvents (ionic melts) . . . . . . . . . .

. . . . . .

10.4.1

Acid-base definitions used for the description of donor-acceptor interactions in ionic media . . . . . . . . . . . . . . . . . . . . . . . . The Lewis definition . . . . . . . . . . . . . . . . . . . . . . . . . . . The Lux-Flood definition. . . . . . . . . . . . . . . . . . . . . . . . . The features of ionic melts as media for acid-base interactions . . . . . Oxygen-less media . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oxygen-containing melts . . . . . . . . . . . . . . . . . . . . . . . . . The effect of the ionic solvent composition on acid-base equilibria . . . Methods for estimations of acidities of solutions based on ionic melts . On studies of the homogeneous acid-base reactions in ionic melts . . . Nitrate melts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sulphate melts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Silicate melts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The equimolar mixture KCl-NaCl . . . . . . . . . . . . . . . . . . . . Other alkaline halide melts . . . . . . . . . . . . . . . . . . . . . . . . Reactions of melts with gaseous acids and bases . . . . . . . . . . . . High-temperature hydrolysis of molten halides . . . . . . . . . . . . . The processes of removal of oxide admixtures from melts . . . . . . . ELECTRONIC AND ELECTRICAL EFFECTS OF SOLVENTS . Theoretical treatment of solvent effects on electronic and vibrational spectra of compounds in condensed media. . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

608 611 612 612 612 616

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

617 617 618 618 619 619 620 623 625 625 627 628 629 631 632 632 633 639

VICTOR CHERGINETS

10.4.1.1 10.4.1.2 10.4.2 10.4.2.1 10.4.2.2 10.4.2.3 10.4.3 10.4.4 10.4.4.1 10.4.4.2 10.4.4.3 10.4.4.4 10.4.4.5 10.4.5 10.4.5.1 10.4.5.2 11 11.1

. . . . 639

MATI KARELSON

11.1.1 11.1.2 11.1.3 11.1.4 11.1.5 11.2

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theoretical treatment of solvent cavity effects on electronic-vibrational spectra of molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theoretical treatment of solvent electrostatic polarization on electronic-vibrational spectra of molecules . . . . . . . . . . . . . . . . . . . Theoretical treatment of solvent dispersion effects on electronic-vibrational spectra of molecules . . . . . . . . . . . . . . . . . . . Supermolecule approach to the intermolecular interactions in condensed media Dielectric solvent effects on the intensity of light absorption and the radiative rate constant . . . . . . . . . . . . . . . . . . . . . . . . . . . .

639 647 649 671 674 680

TAI-ICHI SHIBUYA

11.2.1 11.2.2 11.2.3 12 12.1

The Chako formula or the Lorentz-Lorenz correction . . . . . . . . . . The generalized local-field factor for the ellipsoidal cavity . . . . . . . Dielectric solvent effect on the radiative rate constant. . . . . . . . . . OTHER PROPERTIES OF SOLVENTS, SOLUTIONS, AND PRODUCTS OBTAINED FROM SOLUTIONS . . . . . . . . Rheological properties, aggregation, permeability, molecular structure, crystallinity, and other properties affected by solvents . . . . . . . . .

. . . . 680 . . . . 680 . . . . 682 . . . . 683 . . . . 683

GEORGE WYPYCH

12.1.1 12.1.2 12.1.3 12.1.4 12.1.5 12.2

Rheological properties . . . . . . . . . . . . . . . . . . . . . Aggregation . . . . . . . . . . . . . . . . . . . . . . . . . . Permeability . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular structure and crystallinity . . . . . . . . . . . . . . Other properties affected by solvents . . . . . . . . . . . . . Chain conformations of polysaccharides in different solvents .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

683 689 693 697 700 706

12.2.1 12.2.2

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 706 Structure and conformation of polysaccharides in solution . . . . . . . . . . . 707

RANIERI URBANI AND ATTILIO CESÀRO

Table of contents

12.2.2.1 12.2.2.2 12.2.3 12.2.4 12.2.4.1 12.2.4.2 12.2.5 12.2.6 12.2.6.1 12.2.6.2 12.2.6.3 12.2.6.4 12.2.7 13 13.1

ix

Chemical structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solution chain conformation . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental evidence of solvent effect on oligosaccharide conformational equilibria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theoretical evaluation of solvent effect on conformational equilibria of sugars Classical molecular mechanics methods . . . . . . . . . . . . . . . . . . . . . Molecular dynamic methods . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent effect on chain dimensions and conformations of polysaccharides . . . Solvent effect on charged polysaccharides and the polyelectrolyte model . . . Experimental behavior of polysaccharides polyelectrolytes . . . . . . . . . . . The Haug and Smidsrød parameter: description of the salt effect on the chain dimension. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The statistical thermodynamic counterion-condensation theory of Manning . . Conformational calculations of charged polysaccharides . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . EFFECT OF SOLVENT ON CHEMICAL REACTIONS AND REACTIVITY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent effects on chemical reactivity . . . . . . . . . . . . . . . . . . . . . .

707 707 711 715 715 720 722 726 726 727 729 731 733 737 737

ROLAND SCHMID

13.1.1 13.1.2 13.1.3 13.1.4 13.1.5 13.1.6 13.1.7

13.1.8 13.2

Introduction. . . . . . . . . . . . . . . . . . . . . . . . The dielectric approach. . . . . . . . . . . . . . . . . . The chemical approach . . . . . . . . . . . . . . . . . . Dielectric vs. chemical approach . . . . . . . . . . . . . Conceptual problems with empirical solvent parameters The physical approach . . . . . . . . . . . . . . . . . . Some highlights of recent investigations . . . . . . . . . The like dissolves like rule . . . . . . . . . . . . . . . . Water’s anomalies . . . . . . . . . . . . . . . . . . . . The hydrophobic effect . . . . . . . . . . . . . . . . . . The structure of liquids . . . . . . . . . . . . . . . . . . Solvent reorganization energy in ET . . . . . . . . . . . The solution ionic radius . . . . . . . . . . . . . . . . . The future of the phenomenological approach . . . . . . Solvent effects on free radical polymerization . . . . . .

13.2.1 13.2.2 13.2.2.1 13.2.2.2 13.2.2.3 13.2.2.4 13.2.3 13.2.3.1 13.2.3.1.1 13.2.3.1.2 13.2.3.1.3 13.2.3.2 13.2.3.2.1 13.2.3.2.2 13.2.3.2.3 13.2.3.3 13.2.3.3.1 13.2.3.3.2 13.2.3.3.3 13.2.3.3.4 13.2.3.4 13.2.3.4.1

Introduction. . . . . . . . . . . . . . . . . . . . . . . Homopolymerization . . . . . . . . . . . . . . . . . . Initiation . . . . . . . . . . . . . . . . . . . . . . . . Propagation . . . . . . . . . . . . . . . . . . . . . . . Transfer . . . . . . . . . . . . . . . . . . . . . . . . . Termination. . . . . . . . . . . . . . . . . . . . . . . Copolymerization. . . . . . . . . . . . . . . . . . . . Polarity effect. . . . . . . . . . . . . . . . . . . . . . Basic mechanism . . . . . . . . . . . . . . . . . . . . Copolymerization model . . . . . . . . . . . . . . . . Evidence for polarity effects in propagation reactions . Radical-solvent complexes . . . . . . . . . . . . . . . Basic mechanism . . . . . . . . . . . . . . . . . . . . Copolymerization model . . . . . . . . . . . . . . . . Experimental evidence . . . . . . . . . . . . . . . . . Monomer-solvent complexes. . . . . . . . . . . . . . Introduction. . . . . . . . . . . . . . . . . . . . . . . Monomer-monomer complex participation model . . . Monomer-monomer complex dissociation model . . . Specific solvent effects . . . . . . . . . . . . . . . . . Bootstrap model . . . . . . . . . . . . . . . . . . . . Basic mechanism . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

737 737 738 742 744 746 753 753 755 758 762 765 768 772 777

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

777 777 777 778 779 779 779 780 780 781 781 782 782 782 783 785 785 785 790 791 791 791

MICHELLE L. COOTE AND THOMAS P. DAVIS

. . . . . . . . . . . . . . . . . . . . . .

x


13.2.3.4.2 13.2.3.4.3 13.2.4 13.3

Copolymerization model . . . . . . . . . . . . . . . . Experimental evidence . . . . . . . . . . . . . . . . . Concluding remarks . . . . . . . . . . . . . . . . . . Effects of organic solvents on phase-transfer catalysis

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

791 793 795 798

13.3.1 13.3.1.1 13.3.1.2 13.3.1.3 13.3.1.4

Two-phase phase-transfer catalytic reactions . . . . . . . . . . . . . . . . . Theoretical analysis of the polarity of the organic solvents and the reactions . Effect of organic solvent on the reaction in various reaction systems . . . . . Effects of the organic solvents on the reactions in other catalysts . . . . . . . Effect of the volume of organic solvent and water on the reactions in various reaction systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effects of organic solvents on other phase-transfer catalytic reactions . . . . Other effects on the phase-transfer catalytic reactions . . . . . . . . . . . . . Three-phase reactions (triphase catalysis) . . . . . . . . . . . . . . . . . . . The interaction between solid polymer (hydrophilicity) and the organic solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effect of solvents on the reaction in triphase catalysis . . . . . . . . . . . . . Effect of volume of organic solvent and water on the reactions in triphase catalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effect of polymerization solvent on the chemical structure and curing of aromatic poly(amideimide). . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

801 801 805 811

. . . .

822 825 828 830

MAW-LING WANG

13.3.1.5 13.3.1.6 13.3.2 13.3.2.1 13.3.2.2 13.3.2.3 13.4

. 830 . 833 . 836 . 841

NORIO TSUBOKAWA

13.4.1 13.4.2 13.4.2.1 13.4.2.2 13.4.3 13.4.3.1 13.4.3.2 13.4.4 14 14.1

Introduction. . . . . . . . . . . . . . . . . . . . . . . . Effect of solvent on the chemical structure of PAI. . . . Imide and amide bond content of PAI . . . . . . . . . . Intrinsic viscosity and carboxyl group content . . . . . . Effect of solvent on the curing of PAI by heat treatment Chemical structure of PAI after heat treatment . . . . . Curing PAI by post-heating . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . SOLVENT USE IN VARIOUS INDUSTRIES . . . . Adhesives and sealants . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

841 842 842 844 844 844 845 846 847 847

14.2

Aerospace. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 852

14.3

Asphalt compounding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 855

14.4 14.4.1

Biotechnology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 856 Organic solvents in microbial production processes . . . . . . . . . . . . . . . 856

14.4.1.1 14.4.1.2 14.4.1.3 14.4.1.4 14.4.2

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . Toxicity of organic solvents . . . . . . . . . . . . . . . . Solvent-tolerant bacteria . . . . . . . . . . . . . . . . . . Biotransformation using solvent-tolerant microorganisms. Solvent-resistant microorganisms . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

856 859 862 863 865

14.4.2.1 14.4.2.2 14.4.2.2.1 14.4.2.2.2 14.4.2.3 14.4.2.3.1 14.4.2.3.2 14.4.2.4 14.4.2.4.1 14.4.3

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Toxicity of solvents for microorganisms . . . . . . . . . . . . . . . . . . . Spectrum of microorganisms and solvents . . . . . . . . . . . . . . . . . . Mechanisms of solvent toxicity for microorganisms. . . . . . . . . . . . . Adaption of microorganisms to solvents - solvent-resistant microorganisms Spectrum of solvent-resistant microorganisms. . . . . . . . . . . . . . . . Adaption mechanisms of microorganisms to solvents . . . . . . . . . . . . Solvents and microorganisms in the environment and industry - examples . Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Choice of solvent for enzymatic reaction in organic solvent. . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

865 865 865 866 867 867 868 869 869 872

GEORGE WYPYCH GEORGE WYPYCH GEORGE WYPYCH

MICHIAKI MATSUMOTO, SONJA ISKEN, JAN A. M. DE BONT

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

TILMAN HAHN, KONRAD BOTZENHART

TSUNEO YAMANE

Table of contents

14.4.3.1 14.4.3.2 14.4.3.3

xi

14.4.3.4 14.4.3.5 14.5

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Classification of organic solvents . . . . . . . . . . . . . . . . . . Influence of solvent parameters on nature of enzymatic reactions in organic media. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Properties of enzymes affected by organic solvents . . . . . . . . . Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . Coil coating. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . 872 . . . . . . 872

14.6

Cosmetics and personal care products . . . . . . . . . . . . . . . . . . . . . . 881

14.7

Dry cleaning - treatment of textiles in solvents . . . . . . . . . . . . . . . . . 883

14.7.1 14.7.1.1 14.7.1.2 14.7.1.3 14.7.1.4 14.7.1.5 14.7.1.6 14.7.1.7 14.7.2 14.7.2.1 14.7.2.2 14.7.2.3 14.7.3 14.7.3.1 14.7.3.2 14.7.3.3 14.8

Dry cleaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . History of dry cleaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basis of dry cleaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Behavior of textiles in solvents and water . . . . . . . . . . . . . . . . . . . . Removal of soiling in dry cleaning. . . . . . . . . . . . . . . . . . . . . . . . Activity of detergents in dry cleaning . . . . . . . . . . . . . . . . . . . . . . Dry cleaning processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Recycling of solvents in dry cleaning . . . . . . . . . . . . . . . . . . . . . . Spotting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spotting in dry cleaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spotting agents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spotting procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Textile finishing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Waterproofing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Milling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Antistatic finishing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electronic industry - CFC-free alternatives for cleaning in electronic industry.

883 883 884 885 886 887 888 890 891 891 891 892 893 893 893 893 894

14.8.1 14.8.2 14.8.2.1 14.8.2.1.1 14.8.2.1.2 14.8.2.1.3 14.8.2.1.4

Cleaning requirements in the electronic industry . . . . . . . . . . . . . . . . Available alternatives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water based systems; advantages and disadvantages . . . . . . . . . . . . . . Cleaning with DI - water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cleaning with alkaline water-based media . . . . . . . . . . . . . . . . . . . . Aqueous-based cleaning agents containing water soluble organic components . Water-based cleaning agents based on MPC® Technology (MPC = Micro Phase Cleaning) . . . . . . . . . . . . . . . . . . . . . . . . . Advantages and disadvantages of aqueous cleaning media . . . . . . . . . . . Semi-aqueous cleaners based on halogen-free solvents, advantages and disadvantages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water insoluble cleaning fluids . . . . . . . . . . . . . . . . . . . . . . . . . Water-soluble, water-based cleaning agents . . . . . . . . . . . . . . . . . . . Comparison of the advantages (+) and disadvantages (-) of semi-aqueous cleaning fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Other solvent based cleaning systems . . . . . . . . . . . . . . . . . . . . . . Cleaning of tools and auxiliaries . . . . . . . . . . . . . . . . . . . . . . . . . Cleaning substrates and contamination. . . . . . . . . . . . . . . . . . . . . . Compatibility of stencil and cleaning agent . . . . . . . . . . . . . . . . . . . Different cleaning media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Comparison of manual cleaning vs. automated cleaning. . . . . . . . . . . . . Cleaning equipment for stencil cleaning applications . . . . . . . . . . . . . . Stencil cleaning in screen printing machines. . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cleaning agents and process technology available for cleaning PCB . . . . . . Flux remove and aqueous process . . . . . . . . . . . . . . . . . . . . . . . . The limits of a no-clean process . . . . . . . . . . . . . . . . . . . . . . . . .

894 896 897 897 898 898

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

873 875 879 880

GEORGE WYPYCH GEORGE WYPYCH KASPAR D. HASENCLEVER

14.8.2.1.5 14.8.2.2 14.8.2.2.1 14.8.2.2.2 14.8.2.2.3 14.8.2.3 14.8.3 14.8.3.1 14.8.3.2 14.8.3.3 14.8.3.4 14.8.3.5 14.8.3.6 14.8.3.7 14.8.4 14.8.4.1 14.8.4.1.1

MARTIN HANEK, NORBERT LÖW, ANDREAS MÜHLBAUER

899 899 900 901 901 901 902 904 904 905 906 908 909 911 911 911 911 911

xii


14.8.4.1.2 14.8.4.1.3 14.8.4.1.4 14.8.4.2 14.8.4.2.1

Different cleaning media and cleaning processes . . . . . . . . . . . . . . . . Semi-aqueous cleaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Aqueous cleaning in spray in air cleaning equipment . . . . . . . . . . . . . . Flux removal from printed circuit boards - water-free cleaning processes . . . Water-free cleaning processes using HFE (hydrofluoroethers) in combination with a cosolvent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.8.4.2.2 Water-free cleaning processes in closed, one-chamber vapor defluxing systems 14.8.5 Criteria for assessment and evaluation of cleaning results . . . . . . . . . . . . 14.8.6 Cost comparison of different cleaning processes. . . . . . . . . . . . . . . . . 14.9 Fabricated metal products . . . . . . . . . . . . . . . . . . . . . . . . . . . .

912 913 913 914 915 916 917 919 920

GEORGE WYPYCH

14.10

Food industry - solvents for extracting vegetable oils . . . . . . . . . . . . . . 923 PHILLIP J. WAKELYN, PETER J. WAN

14.10.1 14.10.2 14.10.2.1 14.10.2.1.1 14.10.2.1.2 14.10.2.1.3 14.10.2.2 14.10.2.2.1 14.10.2.2.2 14.10.2.2.3 14.10.2.2.4

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regulatory concerns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Workplace regulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Air Contaminants Standard (29 CFR 1910.1000) . . . . . . . . . . . . . . . . Hazard Communication Standard (HCS) (29 CFR 1910.1200) . . . . . . . . . Process Safety Management (PSM) Standard (29 CFR 1910.119) . . . . . . . Environmental regulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . Clean Air Act (CAA; 42 U.S. Code 7401 et seq.) . . . . . . . . . . . . . . . . Clean Water Act (CWA; 33 U.S. Code 1251 et seq.) . . . . . . . . . . . . . . Resource Conservation and Recovery Act (RCRA; 42 U.S.Code 6901 et seq.) . Emergency Planning and Community Right-to-Know Act (EPCRA; 42 U.S. Code 11001 et seq.) . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.10.2.2.5 Toxic Substances Control Act (TSCA; 15 U.S. Code 2601 et seq.) . . . . . . . 14.10.2.3 Food safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.10.3 The solvent extraction process . . . . . . . . . . . . . . . . . . . . . . . . . . 14.10.3.1 Preparation for extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.10.3.2 Oil extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.10.3.3 Processing crude oil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.10.4 Review of solvents studied for extraction efficiency. . . . . . . . . . . . . . . 14.10.4.1 Hydrocarbon solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.10.4.1.1 Nomenclature, structure, composition and properties of hydrocarbons . . . . . 14.10.4.1.2 Performance of selected hydrocarbon solvents. . . . . . . . . . . . . . . . . . 14.10.5 Future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.11 Ground transportation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

923 924 925 925 926 927 927 929 932 932 933 933 934 935 936 938 938 940 941 942 942 946 950

GEORGE WYPYCH

14.12

Inorganic chemical industry . . . . . . . . . . . . . . . . . . . . . . . . . . . 950

14.13

Iron and steel industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 951

14.14

Lumber and wood products - Wood preservation treatment: significance of solvents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 953

14.14.1 14.14.2 14.14.2.1 14.14.2.2 14.15

General aspects . . . . . . . . . . . . Role of solvents . . . . . . . . . . . Occurrence . . . . . . . . . . . . . . Technical and environmental aspects Medical applications . . . . . . . . .

14.16

Metal casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 957

14.17

Motor vehicle assembly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 958

14.18

Organic chemical industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 962

GEORGE WYPYCH GEORGE WYPYCH

TILMAN HAHN, KONRAD BOTZENHART, FRITZ SCHWEINSBERG, GERHARD VOLLAND

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

953 954 954 955 955

GEORGE WYPYCH GEORGE WYPYCH GEORGE WYPYCH GEORGE WYPYCH

Table of contents

xiii

14.19 14.19.1

Paints and coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 963 Architectural surface coatings and solvents . . . . . . . . . . . . . . . . . . . 963

14.19.1.1 14.19.1.2 14.19.2

General aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 963 Technical aspects and properties of coating materials . . . . . . . . . . . . . . 963 Recent advances in coalescing solvents for waterborne coatings . . . . . . . . 969


DAVID RANDALL

14.19.2.1 Introduction. . . . . . . . . . . . . . . . . . . . 14.19.2.2 Water based coatings . . . . . . . . . . . . . . . 14.19.2.3 Emulsion polymers . . . . . . . . . . . . . . . . 14.19.2.4 Role of a coalescing solvent . . . . . . . . . . . 14.19.2.5 Properties of coalescing agents. . . . . . . . . . 14.19.2.5.1 Hydrolytic stability . . . . . . . . . . . . . . . . 14.19.2.5.2 Water solubility. . . . . . . . . . . . . . . . . . 14.19.2.5.3 Freezing point . . . . . . . . . . . . . . . . . . 14.19.2.5.4 Evaporation rate . . . . . . . . . . . . . . . . . 14.19.2.5.5 Odor . . . . . . . . . . . . . . . . . . . . . . . 14.19.2.5.6 Color . . . . . . . . . . . . . . . . . . . . . . . 14.19.2.5.7 Coalescing efficiency. . . . . . . . . . . . . . . 14.19.2.5.8 Incorporation . . . . . . . . . . . . . . . . . . . 14.19.2.5.9 Improvement of physical properties . . . . . . . 14.19.2.5.10 Biodegradability. . . . . . . . . . . . . . . . . 14.19.2.5.11 Safety . . . . . . . . . . . . . . . . . . . . . . 14.19.2.6 Comparison of coalescing solvents. . . . . . . . 14.19.2.7 Recent advances in diester coalescing solvents . 14.19.2.8 Appendix - Classification of coalescing solvents 14.20 Petroleum refining industry . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . .

969 970 970 971 972 972 972 972 972 972 973 973 973 973 973 973 973 974 975 975

GEORGE WYPYCH

14.21 14.21.1

Pharmaceutical industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 977 Use of solvents in the manufacture of drug substances (DS) and drug products (DP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 977 MICHEL BAUER, CHRISTINE BARTHÉLÉMY

14.21.1.1 14.21.1.2 14.21.1.2.1 14.21.1.2.2 14.21.1.3

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Where are solvents used in the manufacture of pharmaceutical drugs? . . . Intermediates of synthesis, DS and excipients . . . . . . . . . . . . . . . . Drug products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Impacts of the nature of solvents and their quality on the physicochemical characteristics of raw materials and DP . . . . . . . . . . . . . . . . . . . 14.21.1.3.1 Raw materials (intermediates, DS, excipients) . . . . . . . . . . . . . . . . 14.21.1.3.2 Drug product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.21.1.3.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.21.1.4 Setting specifications for solvents . . . . . . . . . . . . . . . . . . . . . . 14.21.1.4.1 Solvents used for the raw material manufacture . . . . . . . . . . . . . . . 14.21.1.4.2 Solvents used for the DP manufacture . . . . . . . . . . . . . . . . . . . . 14.21.1.5 Quality of solvents and analysis . . . . . . . . . . . . . . . . . . . . . . . 14.21.1.5.1 Quality of solvents used in spectroscopy. . . . . . . . . . . . . . . . . . . 14.21.1.5.2 Quality of solvents used in chromatography . . . . . . . . . . . . . . . . . 14.21.1.5.3 Quality of solvents used in titrimetry . . . . . . . . . . . . . . . . . . . . 14.21.1.6 Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.21.2 Predicting cosolvency for pharmaceutical and environmental applications .

. . . .

. . . .

977 979 979 984

. . . . . . . . . . . . .

. . . . . . . . . . . . .

985 985 988 989 990 990 991 991 991 993 996 996 997

. . . . . .

. . . . . .

997 998 1000 1001 1003 1007

AN LI

14.21.2.1 14.21.2.2 14.21.2.3 14.21.2.4 14.21.2.5 14.21.2.6

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications of cosolvency in pharmaceutical sciences and industry . . Applications of cosolvency in environmental sciences and engineering. Experimental observations . . . . . . . . . . . . . . . . . . . . . . . . Predicting cosolvency in homogeneous liquid systems . . . . . . . . . Predicting cosolvency in non-ideal liquid mixtures . . . . . . . . . . .

. . . . . .

. . . . . .

xiv


14.21.2.7 14.22

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1013 Polymers and man-made fibers. . . . . . . . . . . . . . . . . . . . . . . . . . 1016

14.23

Printing industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1020

14.24

Pulp and paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1023

14.25

Rubber and plastics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1025

14.26

Use of solvents in the shipbuilding and ship repair industry . . . . . . . . . . . 1026

14.26.1 14.26.2 14.26.3 14.26.4 14.26.4.1 14.26.4.2 14.26.4.3 14.26.5 14.26.6 14.26.7 14.26.8 14.26.9 14.26.10 14.27

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . Shipbuilding and ship repair operations . . . . . . . . . . . Coating operations . . . . . . . . . . . . . . . . . . . . . . Cleaning operations using organic solvents . . . . . . . . . Surface preparation and initial corrosion protection . . . . . Cleaning operations after coatings are applied . . . . . . . . Maintenance cleaning of equipment items and components . Marine coatings. . . . . . . . . . . . . . . . . . . . . . . . Thinning of marine coatings . . . . . . . . . . . . . . . . . Solvent emissions . . . . . . . . . . . . . . . . . . . . . . Solvent waste . . . . . . . . . . . . . . . . . . . . . . . . . Reducing solvent usage, emissions, and waste. . . . . . . . Regulations and guidelines for cleaning solvents . . . . . . Stone, clay, glass, and concrete . . . . . . . . . . . . . . .

14.28

Textile industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1041

14.29

Transportation equipment cleaning. . . . . . . . . . . . . . . . . . . . . . . . 1042

14.30

Water transportation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1042

14.31

Wood furniture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1043

14.32 15 15.1

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1045 METHODS OF SOLVENT DETECTION AND TESTING . . . . . . . . . 1053 Standard methods of solvent analysis . . . . . . . . . . . . . . . . . . . . . . 1053

15.1.1 15.1.2 15.1.3 15.1.4 15.1.5 15.1.6 15.1.7 15.1.8 15.1.9 15.1.10 15.1.11 15.1.12 15.1.13 15.1.14 15.1.15 15.1.16 15.1.17 15.1.18 15.1.19

Alkalinity and acidity. . . . . . . . . . Autoignition temperature . . . . . . . . Biodegradation potential . . . . . . . . Boiling point . . . . . . . . . . . . . . Bromine index . . . . . . . . . . . . . Calorific value . . . . . . . . . . . . . Cleaning solvents . . . . . . . . . . . . Color . . . . . . . . . . . . . . . . . . Corrosion (effect of solvents) . . . . . Density . . . . . . . . . . . . . . . . . Dilution ratio . . . . . . . . . . . . . . Dissolving and extraction . . . . . . . Electric properties . . . . . . . . . . . Environmental stress crazing . . . . . . Evaporation rate . . . . . . . . . . . . Flammability limits. . . . . . . . . . . Flash point . . . . . . . . . . . . . . . Freezing point . . . . . . . . . . . . . Free halogens in halogenated solvents .

GEORGE WYPYCH GEORGE WYPYCH GEORGE WYPYCH GEORGE WYPYCH MOHAMED SERAGELDIN, DAVE REEVES

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

1026 1026 1026 1027 1027 1028 1031 1031 1032 1033 1035 1036 1037 1039

GEORGE WYPYCH GEORGE WYPYCH GEORGE WYPYCH GEORGE WYPYCH GEORGE WYPYCH

GEORGE WYPYCH

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

1053 1054 1054 1055 1055 1056 1056 1056 1057 1057 1057 1058 1058 1059 1059 1059 1060 1061 1061

Table of contents

xv

15.1.20 15.1.21 15.1.22 15.1.23 15.1.24 15.1.25 15.1.26 15.1.27 15.1.28 15.1.29 15.1.30 15.1.31 15.1.32 15.1.33 15.1.34 15.1.35 15.2 15.2.1

Gas chromatography . . . . . . . . . . . . . . . . . . . . . . . . . . . . Labeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Odor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Paints standards related to solvents . . . . . . . . . . . . . . . . . . . . pH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Purity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Refractive index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Residual solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solubility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent partitioning in soils . . . . . . . . . . . . . . . . . . . . . . . . Solvent extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Specifications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sustained burning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vapor pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Volatile organic compound content, VOC . . . . . . . . . . . . . . . . . Special methods of solvent analysis . . . . . . . . . . . . . . . . . . . . Use of breath monitoring to assess exposures to volatile organic solvents

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . .

1061 1062 1062 1063 1063 1063 1066 1066 1066 1066 1067 1067 1067 1068 1068 1069 1078 1078

15.2.1.1 15.2.1.2 15.2.1.3 15.2.1.3.1 15.2.1.3.2 15.2.1.3.3 15.2.1.4 15.2.1.5 15.2.1.6 15.2.1.7 15.2.2

Principles of breath monitoring . . . . . . . . . Types of samples used for biological monitoring Fundamentals of respiratory physiology . . . . . Ventilation . . . . . . . . . . . . . . . . . . . . Partition coefficients . . . . . . . . . . . . . . . Gas exchange . . . . . . . . . . . . . . . . . . . Types of exhaled air samples. . . . . . . . . . . Breath sampling methodology . . . . . . . . . . When is breath monitoring appropriate? . . . . . Examples of breath monitoring. . . . . . . . . . A simple test to determine toxicity using bacteria

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

1078 1080 1080 1081 1081 1082 1083 1084 1087 1088 1095

15.2.2.1 15.2.2.2 15.2.2.3 15.2.2.4 15.2.2.5 15.2.2.6 15.2.2.7 15.2.2.8 15.2.2.9 15.2.2.10 15.2.3

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Toxicity defined . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . An alternative. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chemicals tested . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Comparisons with other tests . . . . . . . . . . . . . . . . . . . . . . . . . Toxic herbicides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Toxicity of divalent cations . . . . . . . . . . . . . . . . . . . . . . . . . Toxicity of organics in the presence of EDTA . . . . . . . . . . . . . . . . Mechanism for reduction of the dye . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Description of an innovative GC method to assess the influence of crystal texture and drying conditions on residual solvent content in pharmaceutical products. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

1095 1095 1097 1099 1103 1107 1108 1108 1110 1111

MYRTO PETREAS

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

JAMES L. BOTSFORD

CHRISTINE BARTHÉLÉMY, MICHEL BAUER

. . 1113

15.2.3.1 15.2.3.2 15.2.3.2.1 15.2.3.2.2 15.2.3.2.3 16 16.1

Description of the RS determination method . . . . . . . . . . . . . . . . . Application: Influence of crystal texture and drying conditions on RS content First example: monocrystalline particles of paracetamol . . . . . . . . . . . Second example: polycrystalline particles of meprobamate and ibuprofen . . Third example: polycrystalline particles of paracetamol. . . . . . . . . . . . RESIDUAL SOLVENTS IN PRODUCTS . . . . . . . . . . . . . . . . . Residual solvents in various products . . . . . . . . . . . . . . . . . . . . .

. . . . . . .

1113 1114 1116 1119 1122 1125 1125

16.2

Residual solvents in pharmaceutical substances . . . . . . . . . . . . . . . . . 1129

16.2.1 16.2.2

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1129 Why should we look for RS? . . . . . . . . . . . . . . . . . . . . . . . . . . . 1129

GEORGE WYPYCH MICHEL BAUER, CHRISTINE BARTHÉLÉMY

xvi

16.2.2.1 16.2.2.2 16.2.2.3 16.2.2.4 16.2.2.5 16.2.2.5.1 16.2.2.5.2 16.2.3 16.2.3.1 16.2.3.2 16.2.3.3 16.2.3.3.1 16.2.3.3.2 16.2.3.3.3 16.2.4 16.2.4.1 16.2.4.2 16.2.4.3 16.2.4.4 16.2.4.5 16.2.4.6 16.2.5 17 17.1


Modifying the acceptability of the drug product . . . . . . . . . . . . . . . Modifying the physico-chemical properties of drug substances (DS) and drug products (DP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Implications of possible drug/container interactions . . . . . . . . . . . . . As a tool for forensic applications . . . . . . . . . . . . . . . . . . . . . . As a source of toxicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . General points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Brief overview of the toxicology of solvents. . . . . . . . . . . . . . . . . How to identify and control RS in pharmaceutical substances?. . . . . . . Loss of weight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Miscellaneous methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . Gas chromatography (GC) . . . . . . . . . . . . . . . . . . . . . . . . . . General points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Review of methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Official GC methods for RS determination . . . . . . . . . . . . . . . . . How to set specifications? Examination of the ICH guidelines for residual solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Classification of residual solvents by risk assessment . . . . . . . . . . . . Definition of PDE. Method for establishing exposure limits . . . . . . . . Limits for residual solvents. . . . . . . . . . . . . . . . . . . . . . . . . . Analytical procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions regarding the ICH guideline . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ENVIRONMENTAL IMPACT OF SOLVENTS . . . . . . . . . . . . . The environmental fate and movement of organic solvents in water, soil, and air . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . 1129 . . . . . . . . . . . . .

. . . . . . . . . . . . .

1130 1131 1131 1131 1131 1132 1133 1133 1133 1134 1134 1135 1139

. . . . . . . . .

. . . . . . . . .

1140 1143 1143 1143 1143 1145 1145 1146 1149

. . 1149

WILLIAM R. ROY

17.1.1 17.1.2 17.1.2.1 17.1.2.2 17.1.2.3 17.1.2.4 17.1.3 17.1.3.1 17.1.3.2 17.1.3.3 17.1.4 17.1.4.1 17.1.4.2 17.1.5 17.1.5.1 17.1.5.2 17.1.5.3 17.1.6 17.1.6.1 17.1.6.2 17.1.6.3 17.1.7 17.2

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solubility . . . . . . . . . . . . . . . . . . . . . . . . . . . . Volatilization . . . . . . . . . . . . . . . . . . . . . . . . . . Degradation. . . . . . . . . . . . . . . . . . . . . . . . . . . Adsorption . . . . . . . . . . . . . . . . . . . . . . . . . . . Soil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Volatilization . . . . . . . . . . . . . . . . . . . . . . . . . . Adsorption . . . . . . . . . . . . . . . . . . . . . . . . . . . Degradation. . . . . . . . . . . . . . . . . . . . . . . . . . . Air . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Degradation. . . . . . . . . . . . . . . . . . . . . . . . . . . Atmospheric residence time . . . . . . . . . . . . . . . . . . The 31 solvents in water . . . . . . . . . . . . . . . . . . . . Solubility . . . . . . . . . . . . . . . . . . . . . . . . . . . . Volatilization from water. . . . . . . . . . . . . . . . . . . . Degradation in water . . . . . . . . . . . . . . . . . . . . . . Soil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Volatilization . . . . . . . . . . . . . . . . . . . . . . . . . . Adsorption . . . . . . . . . . . . . . . . . . . . . . . . . . . Degradation. . . . . . . . . . . . . . . . . . . . . . . . . . . Air . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fate-based management of organic solvent-containing wastes

17.2.1 17.2.1.1 17.2.1.2 17.2.1.3 17.2.1.4

Introduction. . . . . . . . . . . . . . . . . . . . . . . The waste disposal site . . . . . . . . . . . . . . . . . The advection-dispersion model and the required input Maximum permissible concentrations . . . . . . . . . Distribution of organic compounds in leachate . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . .

1149 1150 1150 1150 1151 1151 1151 1151 1152 1153 1153 1153 1154 1154 1154 1155 1155 1157 1157 1159 1160 1161 1162

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

1162 1163 1164 1164 1164

WILLIAM R. ROY

. . . . .

. . . . .

. . . . .

. . . . .

Table of contents

xvii

17.2.2 17.2.3 17.3

Movement of solvents in groundwater . . . . . . . . . . . . . . . . . . . . . . 1166 Mass limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1167 Environmental fate and ecotoxicological effects of glycol ethers . . . . . . . . 1169

17.3.1 17.3.2 17.3.3 17.3.4 17.3.4.1 17.3.4.2 17.3.5 17.4

Introduction. . . . . . . . . . . . . . . . . . . . . . . Occurrence . . . . . . . . . . . . . . . . . . . . . . . Environmental behavior . . . . . . . . . . . . . . . . Ecotoxicity . . . . . . . . . . . . . . . . . . . . . . . Survival and growth . . . . . . . . . . . . . . . . . . Reproduction and development . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . Organic solvent impacts on tropospheric air pollution .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

1169 1170 1171 1175 1175 1185 1187 1188

17.4.1 17.4.2 17.4.2.1 17.4.2.2 17.4.2.3 17.4.2.3.1 17.4.2.3.2 17.4.2.3 17.4.2.4 17.4.3 17.4.3.1 17.4.3.2 17.4.3.3 17.4.3.3.1 17.4.4 17.4.5 18

Sources and impacts of volatile solvents . . . . . . . . . . . . . . . . . . Modes and scales of impact . . . . . . . . . . . . . . . . . . . . . . . . Direct exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Formation of secondary compounds . . . . . . . . . . . . . . . . . . . . Spatial scales of secondary effects . . . . . . . . . . . . . . . . . . . . . Global impacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stratospheric ozone depletion . . . . . . . . . . . . . . . . . . . . . . . Global climate forcing . . . . . . . . . . . . . . . . . . . . . . . . . . . Urban and regional scales . . . . . . . . . . . . . . . . . . . . . . . . . Tropospheric ozone. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tropospheric photochemistry and ozone formation . . . . . . . . . . . . Assessing solvent impacts on ozone and VOC reactivity . . . . . . . . . Quantification of solvent emissions on ozone formation . . . . . . . . . Regulatory approaches to ozone control and solvents . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CONCENTRATION OF SOLVENTS IN VARIOUS INDUSTRIAL ENVIRONMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . Measurement and estimation of solvents emission and odor. . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . .

1188 1189 1189 1190 1190 1190 1191 1191 1192 1192 1192 1193 1195 1196 1198 1299

JAMES DEVILLERS, AURÉLIE CHEZEAU, ANDRÉ CICOLELLA, ERIC THYBAUD

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

MICHELLE BERGIN, ARMISTEAD RUSSELL

18.1

. . . 1201 . . . 1201

MARGOT SCHEITHAUER

18.1.1 18.1.2 18.1.2.1 18.1.2.2 18.1.2.3 18.1.3 18.1.3.1 18.1.3.2 18.1.3.3 18.1.3.3.1 18.1.3.3.2 18.1.3.4 18.1.4 18.1.4.1 18.1.4.2 18.1.4.2.1 18.1.4.2.2 18.1.4.3 18.1.4.3.1 18.1.4.3.2 18.1.5

Definition “solvent” and “volatile organic compounds” (VOC) . . . . . . Review of sources of solvent emissions . . . . . . . . . . . . . . . . . . Causes for emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . Emissions of VOCs from varnishes and paints . . . . . . . . . . . . . . VOC emissions from emulsion paints . . . . . . . . . . . . . . . . . . . Measuring of VOC-content in paints and varnishes . . . . . . . . . . . . Definition of low-emissive coating materials . . . . . . . . . . . . . . . Determination of the VOC content according to ASTM D 3960-1 . . . . Determination of the VOC content according to ISO/DIS 11 890/1 and 2 VOC content > 15% . . . . . . . . . . . . . . . . . . . . . . . . . . . . VOC content > 0.1 and < 15 %. . . . . . . . . . . . . . . . . . . . . . . Determination of VOC-content in water-thinnable emulsion paints (in-can VOC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Measurement of solvent emissions in industrial plants . . . . . . . . . . Plant requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The determination of the total carbon content in mg C/Nm³. . . . . . . . Flame ionization detector (FID) . . . . . . . . . . . . . . . . . . . . . . Silica gel approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Qualitative and quantitative assessment of individual components in the exhaust-gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Indicator tubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quantitative solvent determination in exhaust gas of plants by means of gas-chromatography . . . . . . . . . . . . . . . . . . . . . . . . . . . . “Odor” definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

1201 1203 1203 1203 1205 1205 1205 1205 1206 1206 1208

. . . . . .

. . . . . .

. . . . . .

1208 1209 1209 1214 1214 1214

. . . 1215 . . . 1215 . . . 1215 . . . 1219

xviii

18.1.6 18.1.6.1 18.1.6.2 18.1.6.3 18.1.6.4


18.2

Measurement of odor in materials and industrial plants . . . . . . . . . . Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Odor determination by means of the “electronic nose” . . . . . . . . . . Odor determination by means of the olfactometer . . . . . . . . . . . . . Example for odor determination for selected materials: Determination of odorant concentration in varnished furniture surfaces . . . . . . . . . . . Example of odor determination in industrial plants: Odor measurement in an industrial varnishing plant. . . . . . . . . . . . . . . . . . . . . . . . Prediction of organic solvents emission during technological processes .

18.2.1 18.2.2 18.2.3 18.2.4 18.2.5 18.2.6 18.2.7 18.2.7.1 18.2.7.2 18.2.7.3 18.2.7.4 18.2.8 18.2.9 18.3

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods of degreasing . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Identification of the emitted compounds . . . . . . . . . . . . . . . . . . Emission of organic solvents during technological processes . . . . . . . Verification of the method . . . . . . . . . . . . . . . . . . . . . . . . . Relationships between emission and technological parameters . . . . . . Laboratory test stand . . . . . . . . . . . . . . . . . . . . . . . . . . . . The influence of temperature on emission . . . . . . . . . . . . . . . . . The influence of air velocity on emission . . . . . . . . . . . . . . . . . The relationship between the mass of solvent on wet parts and emissions Emission of solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . Verification in industrial conditions . . . . . . . . . . . . . . . . . . . . Indoor air pollution by solvents contained in paints and varnishes . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

1227 1227 1228 1228 1228 1230 1231 1231 1231 1232 1232 1232 1232 1234

18.3.1 18.3.2 18.3.2.1 18.3.2.2 18.3.3 18.3.3.1 18.3.3.2 18.3.4 18.3.4.1 18.3.4.2 18.3.4.2.1 18.3.4.2.2 18.3.4.2.3 18.3.4.2.4 18.3.4.2.5 18.3.4.2.6 18.3.4.2.7 18.3.5 18.3.5.1 18.3.5.1.1 18.3.5.1.2 18.3.5.2 18.3.5.2.1 18.3.5.2.2 18.4

Composition - solvents in paints and varnishes. Theoretical aspects Occurrence of solvents in paints and varnishes . . . . . . . . . . . Solvents in products . . . . . . . . . . . . . . . . . . . . . . . . . Paints and varnishes . . . . . . . . . . . . . . . . . . . . . . . . . Emission of solvents . . . . . . . . . . . . . . . . . . . . . . . . . Emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Immission. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effects on health of solvents from paints and varnishes . . . . . . . Exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Health effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Toxic responses of skin and mucose membranes . . . . . . . . . . Neurological disorders . . . . . . . . . . . . . . . . . . . . . . . . Carcinogenic effects . . . . . . . . . . . . . . . . . . . . . . . . . Respiratory effects . . . . . . . . . . . . . . . . . . . . . . . . . . Toxic responses of blood . . . . . . . . . . . . . . . . . . . . . . . Toxic responses of the reproductive system . . . . . . . . . . . . . Toxic responses of other organ systems . . . . . . . . . . . . . . . Methods for the examination of solvents in paints and varnishes . . Environmental monitoring . . . . . . . . . . . . . . . . . . . . . . Solvents in products . . . . . . . . . . . . . . . . . . . . . . . . . Emission of solvents . . . . . . . . . . . . . . . . . . . . . . . . . Biological monitoring of solvents in human body fluids . . . . . . Solvents and metabolites in human body fluids and tissues . . . . . Biomarkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent uses with exposure risks . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

1234 1235 1235 1237 1240 1240 1242 1243 1243 1243 1243 1244 1245 1246 1247 1247 1247 1248 1248 1248 1248 1248 1248 1248 1251

18.4.1 18.4.2 18.4.3 18.4.4 18.4.5 18.4.6

Introduction. . . . . . . . . . . . . . . . . . . Exposure assessment . . . . . . . . . . . . . . Production of paints and printing inks . . . . . Painting . . . . . . . . . . . . . . . . . . . . . Printing . . . . . . . . . . . . . . . . . . . . . Degreasing, press cleaning and paint removal .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

1251 1252 1255 1256 1257 1258

18.1.6.5

. . . .

. . . .

. . . .

1222 1222 1222 1223

. . . 1223 . . . 1225 . . . 1227

KRZYSZTOF M. BENCZEK, JOANNA KURPIEWSKA


pentti kalliokoski, kai savolinen

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

Table of contents

xix

18.4.7 18.4.8 18.4.9 18.4.10 18.4.11 19

Dry cleaning . . . . . . . . Reinforced plastics industry Gluing . . . . . . . . . . . Other . . . . . . . . . . . . Summary . . . . . . . . . . REGULATIONS . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

1260 1261 1262 1262 1263 1267

19.1 19.2 19.2.1 19.2.1.1 19.2.1.2

. . . .

. . . .

. . . .

1267 1282 1282 1282

19.2.1.3 19.2.1.4 19.2.1.5 19.3 19.3.1 19.3.1.1 19.3.1.2 19.3.1.3 19.3.2 19.3.2.1 19.3.2.2 19.4 19.4.1 19.4.1.1 19.4.1.2 19.5 19.5.1 19.5.1.1 19.5.1.2 19.5.2 19.5.2.1 19.5.2.2 19.6 19.6.1 19.6.1.1 19.6.1.2 19.6.1.3 19.7 19.7.1 19.7.2 19.8 19.9 19.10

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Air laws and regulations . . . . . . . . . . . . . . . . . . . . . . . . . . Clean Air Act Amendments of 1990 . . . . . . . . . . . . . . . . . . . . Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Title I - Provisions for Attainment and Maintenance of National Ambient Air Quality Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . Title III - Hazardous Air Pollutants . . . . . . . . . . . . . . . . . . . . Title V - Permits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Title VI - Stratospheric Ozone Protection . . . . . . . . . . . . . . . . . Water laws and regulations. . . . . . . . . . . . . . . . . . . . . . . . . Clean Water Act . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effluent Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . Permit Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Safe Drinking Water Act . . . . . . . . . . . . . . . . . . . . . . . . . . Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . National Primary Drinking Water Regulations. . . . . . . . . . . . . . . Land laws & regulations . . . . . . . . . . . . . . . . . . . . . . . . . . Resource Conservation and Recovery Act (RCRA) . . . . . . . . . . . . Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . RCRA, Subtitle C - Hazardous Waste . . . . . . . . . . . . . . . . . . . Multimedia laws and regulations . . . . . . . . . . . . . . . . . . . . . . Pollution Prevention Act of 1990 . . . . . . . . . . . . . . . . . . . . . Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Source Reduction Provisions . . . . . . . . . . . . . . . . . . . . . . . . Toxic Substances Control Act . . . . . . . . . . . . . . . . . . . . . . . Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Controlling toxic substances . . . . . . . . . . . . . . . . . . . . . . . . Occupational laws and regulations . . . . . . . . . . . . . . . . . . . . . Occupational Safety and Health Act . . . . . . . . . . . . . . . . . . . . Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Air contaminants exposure limits . . . . . . . . . . . . . . . . . . . . . Hazard Communication Standard . . . . . . . . . . . . . . . . . . . . . International perspective . . . . . . . . . . . . . . . . . . . . . . . . . . Canada . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . European Union . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tools and resources for solvents . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regulations in Europe . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1284 1288 1292 1292 1293 1293 1293 1293 1294 1294 1294 1295 1295 1295 1295 1296 1297 1297 1297 1298 1300 1300 1300 1301 1301 1301 1301 1302 1302 1303 1303 1304 1306 1311

19.10.1 19.10.2 20 20.1

EEC regulations . . . . . . . . . . . . . . . . . . German regulations. . . . . . . . . . . . . . . . . TOXIC EFFECTS OF SOLVENT EXPOSURE Toxicokinetics, toxicodynamics, and toxicology .

20.1.1 20.1.1.1 20.1.1.2 20.1.1.2.1 20.1.1.2.2

Toxicokinetics and toxicodynamics Exposure . . . . . . . . . . . . . . Uptake . . . . . . . . . . . . . . . Inhalation . . . . . . . . . . . . . . Dermal uptake . . . . . . . . . . .

CARLOS M. NU~ NEZ

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .


. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

1311 1312 1315 1315

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

1315 1315 1315 1316 1316


. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

xx

20.1.1.2 20.1.1.3 20.1.2 20.1.2.1 20.1.2.2 20.1.2.3 20.1.2.4 20.1.2.5 20.1.2.6 20.1.3 20.2


Metabolism, distribution, excretion . . . . . . . . . . . . . . . . . . Modeling of toxicokinetics and modifying factors. . . . . . . . . . . Toxicology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . General effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Specific non-immunological effects . . . . . . . . . . . . . . . . . . Immunological effects . . . . . . . . . . . . . . . . . . . . . . . . . Toxic effects of solvents on other organisms . . . . . . . . . . . . . Carcinogenicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Risk assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cognitive and psychosocial outcome of chronic occupational solvent neurotoxicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

1317 1317 1318 1318 1318 1319 1320 1320 1323 1323

. . . . . 1326

JENNI A OGDEN

20.2.1 20.2.2 20.2.3 20.2.4 20.2.5 20.2.6 20.2.7 20.3

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acute symptoms of solvent neurotoxicity . . . . . . . . . . . . . Categorization of OSN . . . . . . . . . . . . . . . . . . . . . . . Assessment of OSN . . . . . . . . . . . . . . . . . . . . . . . . Do the symptoms of Type 2 OSN resolve? . . . . . . . . . . . . Individual differences in susceptibility to OSN . . . . . . . . . . Psychosocial consequences of OSN, and rehabilitation . . . . . . Pregnancy outcome following maternal organic solvent exposure

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

1326 1327 1327 1328 1330 1331 1331 1333

20.3.1 20.3.2 20.3.3

20.3.6

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Animal studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pregnancy outcome following maternal organic solvent exposure: a meta-analysis of epidemiologic studies . . . . . . . . . . . . . . . . Pregnancy outcome following gestational exposure to organic solvents: a prospective controlled study . . . . . . . . . . . . . . . . . . . . . . A proactive approach for the evaluation of fetal safety in chemical industries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Overall conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . .

20.4

Industrial solvents and kidney disease . . . . . . . . . . . . . . . . . . . . . . 1355

20.4.1 20.4.2 20.4.3 20.4.4 20.4.5 20.4.6 20.5

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental animal studies . . . . . . . . . . . . . . . . . . . . . . . Case reports. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Case control studies . . . . . . . . . . . . . . . . . . . . . . . . . . . Epidemiological assessment . . . . . . . . . . . . . . . . . . . . . . . Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lymphohematopoietic study of workers exposed to benzene including multiple myeloma, lymphoma and chronic lymphatic leukemia. . . . .

KRISTEN I. MCMARTIN, GIDEON KOREN

20.3.4 20.3.5

. . . . 1333 . . . . 1334 . . . . 1338 . . . . 1345 . . . . 1347 . . . . 1353

NACHMAN BRAUTBAR

. . . . . .

. . . . . .

. . . . . .

. . . . . .

1355 1356 1356 1357 1360 1361

. . . . 1363

NACHMAN BRAUTBAR

20.5.1 20.5.2 20.5.3 20.5.4 20.5.5 20.5.6 20.5.7 20.5.8 20.5.9 20.5.10 20.6

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . Routes of exposure . . . . . . . . . . . . . . . . . . . . . . Hematopoietic effects of benzene . . . . . . . . . . . . . . Carcinogenic effects of benzene . . . . . . . . . . . . . . . Risk assessment estimates . . . . . . . . . . . . . . . . . . Levels of exposure . . . . . . . . . . . . . . . . . . . . . . Cell types: hematolymphoproliferative effects of benzene . Epidemiological studies . . . . . . . . . . . . . . . . . . . Solvents and benzene. . . . . . . . . . . . . . . . . . . . . Genetic fingerprint theory . . . . . . . . . . . . . . . . . . Chromosomal aberrations and sister chromatoid exchanges .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

1363 1363 1365 1365 1367 1367 1369 1369 1370 1372 1375

20.7

Hepatotoxicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1379

NACHMAN BRAUTBAR NACHMAN BRAUTBAR

Table of contents

xxi

20.7.1 20.7.2 20.7.3 20.7.4 20.7.5 20.7.6 20.7.7 20.7.8 20.7.9 20.7.10 20.7.11 20.8

Introduction. . . . . . . . . . . . . . . . . . . . . . Individual variability and hepatotoxicity of solvents Non-halogenated solvents . . . . . . . . . . . . . . Solvent mixtures . . . . . . . . . . . . . . . . . . . Trichloroethylene. . . . . . . . . . . . . . . . . . . Tetrachloroethylene . . . . . . . . . . . . . . . . . Toluene . . . . . . . . . . . . . . . . . . . . . . . . Dichloromethane . . . . . . . . . . . . . . . . . . . Stoddard solvent . . . . . . . . . . . . . . . . . . . 1,1,1-Trichloroethane. . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . Solvents and the liver. . . . . . . . . . . . . . . . .

20.8.1 20.8.1.1 20.8.1.2

Normal anatomic and physiologic function of the liver . . . . . . . . Factors influencing solvent hepatotoxicity . . . . . . . . . . . . . . . Microscopic, biochemical and clinical findings associated with liver injury due to solvents . . . . . . . . . . . . . . . . . . . . . . . . . . Hepatotoxicity associated with specific solvents. . . . . . . . . . . . Haloalkanes and haloalkenes . . . . . . . . . . . . . . . . . . . . . . Carbon tetrachloride . . . . . . . . . . . . . . . . . . . . . . . . . . Chloroform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dichloromethane . . . . . . . . . . . . . . . . . . . . . . . . . . . . Trichloroethanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1,1,2,2-Tetrachloroethane . . . . . . . . . . . . . . . . . . . . . . . Tetrachloroethylene and trichloroethylene . . . . . . . . . . . . . . . Other halogenated hydrocarbons . . . . . . . . . . . . . . . . . . . . Styrene and aromatic hydrocarbons . . . . . . . . . . . . . . . . . . N-substituted amides . . . . . . . . . . . . . . . . . . . . . . . . . . Nitroparaffins. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Other solvents and mixed solvents . . . . . . . . . . . . . . . . . . . Toxicity of environmental solvent exposure for brain, lung and heart.

20.8.2 20.8.2.1 20.8.2.2 20.8.2.3 20.8.2.4 20.8.2.5 20.8.2.6 20.8.2.7 20.8.2.8 20.8.2.9 20.8.2.10 20.8.2.11 20.8.2.12 20.9

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

DAVID K. BONAUTO, C. ANDREW BRODKIN, WILLIAM O. ROBERTSON

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

1379 1384 1385 1386 1387 1388 1388 1389 1389 1389 1390 1393

. . . . . 1393 . . . . . 1394 . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

1394 1395 1396 1396 1397 1398 1398 1398 1399 1399 1399 1400 1400 1401 1404

KAYE H. KILBURN

21 21.1

SUBSTITUTION OF SOLVENTS BY SAFER PRODUCTS AND PROCESSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1419 Supercritical solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1419

21.1.1 21.1.1.1 21.1.1.2 21.1.1.3 21.1.2 21.1.2.1 21.1.2.1.1 21.1.2.1.2 21.1.2.1.3 21.1.2.2 21.1.2.2.1 21.1.2.2.2 21.1.2.2.3 21.1.2.2.4 21.1.2.3 21.1.2.4 21.1.2.5 21.1.2.6 21.1.2.7 21.1.3 21.1.3.1

Introduction. . . . . . . . . . . . . . . . . . . . . . . A promising path to green chemistry. . . . . . . . . . Unique and tunable physico-chemical properties . . . Sustainable applications in many different areas. . . . Fundamentals . . . . . . . . . . . . . . . . . . . . . . Phase behavior with supercritical solvents . . . . . . . Experimental methods . . . . . . . . . . . . . . . . . Computational aspects . . . . . . . . . . . . . . . . . Modeling . . . . . . . . . . . . . . . . . . . . . . . . Transport properties of supercritical solvents . . . . . Viscosity . . . . . . . . . . . . . . . . . . . . . . . . Diffusivity . . . . . . . . . . . . . . . . . . . . . . . Thermal conductivity . . . . . . . . . . . . . . . . . . Surface tension . . . . . . . . . . . . . . . . . . . . . Entrainer (co-solvent effects) of supercritical solvents Reaction rate implication in supercritical solvents . . . Sorption behavior of supercritical solvents. . . . . . . Swelling with supercritical solvents . . . . . . . . . . Surfactants and micro-emulsions. . . . . . . . . . . . Separation with supercritical solvents . . . . . . . . . Leaching - generic application . . . . . . . . . . . . .

AYDIN K. SUNOL, SERMIN G. SUNOL

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

1419 1422 1422 1422 1423 1423 1426 1428 1429 1431 1431 1432 1433 1435 1435 1436 1437 1437 1438 1438 1441

xxii


21.1.3.2 21.1.3.3 21.1.3.4 21.1.4 21.1.4.1 21.1.4.1.1 21.1.4.1.2 21.1.4.1.3 21.1.4.2 21.1.4.2.1 21.1.4.2.2 21.1.4.3 21.1.4.4 21.1.4.5 21.1.4.6 21.1.4.7 21.1.4.8 21.1.4.9 21.1.4.10 21.1.4.11 21.1.4.12 21.2

Extraction - generic applications . . . . . . . . . . . . . . . . . . Crystallization - generic applications. . . . . . . . . . . . . . . . Sorption - generic applications . . . . . . . . . . . . . . . . . . . Reactions in supercritical solvents . . . . . . . . . . . . . . . . . Homogenous reactions in supercritical solvents - examples . . . . Homogeneous reactions catalyzed by organometallic compounds Homogeneous reactions of supercritical water . . . . . . . . . . . Homogeneous non-catalytic reactions in supercritical solvents . . Heterogeneous reactions in supercritical solvents - examples . . . Heterogeneous catalytic reactions in supercritical solvents . . . . Heterogeneous non-catalytic reactions in supercritical solvents . . Biochemical reactions - examples . . . . . . . . . . . . . . . . . Polymerization reactions - examples . . . . . . . . . . . . . . . . Materials processing with supercritical solvents . . . . . . . . . . Particle synthesis - generic application. . . . . . . . . . . . . . . Encapsulation - generic application . . . . . . . . . . . . . . . . Spraying and coating - generic application. . . . . . . . . . . . . Extrusion - generic application . . . . . . . . . . . . . . . . . . . Perfusion (impregnation) - generic application . . . . . . . . . . Parts cleaning - generic application . . . . . . . . . . . . . . . . Drying - generic application . . . . . . . . . . . . . . . . . . . . Ionic liquids . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

1442 1443 1443 1444 1445 1446 1447 1448 1448 1449 1450 1451 1451 1452 1453 1454 1454 1454 1454 1455 1455 1459

21.2.1 21.2.2 21.2.2.1 21.2.2.2 21.2.3 21.2.3.1 21.2.3.2 21.2.4 21.2.4.1 21.2.5 21.2.5.1 21.2.5.2 21.2.6 21.3

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fundamental principles of the formation of room temperature ionic liquids Development of ionic liquids. . . . . . . . . . . . . . . . . . . . . . . . . Binary ionic liquid systems. . . . . . . . . . . . . . . . . . . . . . . . . . Catalysis in ionic liquids . . . . . . . . . . . . . . . . . . . . . . . . . . . Reactions involving first generation chloroaluminate(III) ionic liquids . . . Reactions in neutral or second generation ionic liquids . . . . . . . . . . . Electrochemical applications . . . . . . . . . . . . . . . . . . . . . . . . . Electrosynthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Physical characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . Viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oxide solubilities in ionic melts . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

1459 1461 1461 1465 1466 1467 1469 1472 1473 1473 1473 1478 1480 1484

21.3.1 21.3.1.1 21.3.1.2 21.3.2 21.3.3 21.3.3.1 21.3.3.2 21.3.3.3 21.3.3.4 21.3.4 21.3.4.1 21.3.4.2 21.3.4.3 21.3.5 21.4

Methods used for solubility estimations in ionic melts . . . . . . . . . . . . Isothermal saturation method. . . . . . . . . . . . . . . . . . . . . . . . . . Potentiometric titration method . . . . . . . . . . . . . . . . . . . . . . . . Oxygen-containing melts . . . . . . . . . . . . . . . . . . . . . . . . . . . . Halide melts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The eutectic mixture KCl-LiCl (0.41:0.59) . . . . . . . . . . . . . . . . . . Molten KCl-NaCl (0.50:0.50) . . . . . . . . . . . . . . . . . . . . . . . . . Other chloride-based melts . . . . . . . . . . . . . . . . . . . . . . . . . . . Other alkaline halides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . On the possibility to predict oxide solubilities on the base of the existing data The estimation of effect of anion . . . . . . . . . . . . . . . . . . . . . . . . The estimation of effect of melt acidity . . . . . . . . . . . . . . . . . . . . The estimation of effect of temperature . . . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Alternative cleaning technologies/drycleaning installations . . . . . . . . . .

. . . . . . . . . . . . . . .

1484 1485 1486 1487 1487 1487 1488 1491 1493 1494 1494 1494 1495 1495 1497

21.4.1 21.4.1.1 21.4.1.2

Drycleaning with liquid carbon dioxide (LCD) . . . . . . . . . . . . . . . . . 1497 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1497 State of the art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1498

D.W. ROONEY, K.R. SEDDON

VICTOR CHERGINETS

KASPAR D. HASENCLEVER

Table of contents

xxiii

21.4.1.3 21.4.1.4 21.4.1.5 21.4.2 21.4.2.1 21.4.2.2 21.4.2.3 21.4.2.4 21.4.3 22 22.1

Process technology . . . . . . . . . . . . . . . . . . . . . . . . . . . Risks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Competition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wet cleaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kreussler textile cleaning system. . . . . . . . . . . . . . . . . . . . Possibilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Adapting to working practices . . . . . . . . . . . . . . . . . . . . . Future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . SOLVENT RECYCLING, REMOVAL, AND DEGRADATION . Absorptive solvent recovery . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

1498 1499 1500 1501 1501 1503 1504 1504 1505 1507 1507

22.1.1 22.1.2 22.1.2.1 22.1.2.2 22.1.2.3 22.1.2.4 22.1.3 22.1.3.1 22.1.3.2 22.1.3.3 22.1.4 22.1.4.1 22.1.4.2 22.1.4.2.1 22.1.4.2.2 22.1.4.2.3 22.1.4.2.4 22.1.4.3 22.1.4.3.1 22.1.4.3.2 22.1.4.3.3 22.1.4.3.4 22.1.4.3.5 22.1.4.3.6 22.1.5 22.1.5.1 22.1.5.2 22.1.5.2.1 22.1.5.2.2 22.1.5.3 22.1.5.4 22.1.5.5 22.1.5.6 22.2

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basic principles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fundamentals of adsorption . . . . . . . . . . . . . . . . . . . . . . . . Adsorption capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dynamic adsorption in adsorber beds . . . . . . . . . . . . . . . . . . . Regeneration of the loaded adsorbents . . . . . . . . . . . . . . . . . . . Commercially available adsorbents . . . . . . . . . . . . . . . . . . . . Activated carbon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular sieve zeolites . . . . . . . . . . . . . . . . . . . . . . . . . . Polymeric adsorbents . . . . . . . . . . . . . . . . . . . . . . . . . . . . Adsorptive solvent recovery systems. . . . . . . . . . . . . . . . . . . . Basic arrangement of adsorptive solvent recovery with steam desorption. Designing solvent recovery systems . . . . . . . . . . . . . . . . . . . . Design basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Adsorber types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regeneration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Safety requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . Special process conditions . . . . . . . . . . . . . . . . . . . . . . . . . Selection of the adsorbent . . . . . . . . . . . . . . . . . . . . . . . . . Air velocity and pressure drop . . . . . . . . . . . . . . . . . . . . . . . Effects of solvent-concentration, adsorption temperature and pressure . . Influence of humidity. . . . . . . . . . . . . . . . . . . . . . . . . . . . Interactions between solvents and activated carbon . . . . . . . . . . . . Activated carbon service life . . . . . . . . . . . . . . . . . . . . . . . . Examples from different industries. . . . . . . . . . . . . . . . . . . . . Rotogravure printing shops. . . . . . . . . . . . . . . . . . . . . . . . . Packaging printing industry . . . . . . . . . . . . . . . . . . . . . . . . Fixed bed adsorption with circulating hot gas desorption . . . . . . . . . Solvent recovery with adsorption wheels . . . . . . . . . . . . . . . . . Viscose industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Refrigerator recycling . . . . . . . . . . . . . . . . . . . . . . . . . . . Petrochemical industry and tank farms. . . . . . . . . . . . . . . . . . . Chemical industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1507 1509 1509 1510 1511 1512 1513 1513 1514 1515 1515 1515 1518 1518 1519 1521 1522 1523 1523 1526 1526 1528 1529 1531 1531 1531 1532 1533 1535 1535 1539 1539 1541 1543

22.2.1 22.2.2 22.2.3

Activated carbon in fluidized bed adsorption method . . . . . . . . Application of molecular sieves . . . . . . . . . . . . . . . . . . . Continuous process for air cleaning using macroporous particles as adsorption agents . . . . . . . . . . . . . . . . . . . . . . . . . . . Solvent recovery from hazardous wastes. . . . . . . . . . . . . . . Halogenated solvent recovery . . . . . . . . . . . . . . . . . . . . Coating process. . . . . . . . . . . . . . . . . . . . . . . . . . . . Tableting process of pharmaceutical products . . . . . . . . . . . . Energy recovery from waste solvent . . . . . . . . . . . . . . . . . Solvent treatment in a paints and coating plant . . . . . . . . . . .

KLAUS-DIRK HENNING

ISAO KIMURA

22.2.4 22.2.5 22.2.5.1 22.2.5.2 22.2.6 22.3

. . . . . . 1543 . . . . . . 1544 . . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

1546 1548 1549 1549 1552 1553 1555

xxiv


DENIS KARGOL

22.4

Application of solar photocatalytic oxidation to VOC-containing airstreams . . 1559 K. A. MAGRINI, A. S. WATT, L. C. BOYD, E. J. WOLFRUM, S. A. LARSON, C. ROTH

G. C. Glatzmaier

22.4.1 22.4.2 22.4.2.1 22.4.2.2 22.4.3 22.4.3.1 22.4.3.2 22.4.4 23 23.1

Solvent degradation by photocatalytic oxidation. . . . . . . . . . . . . . . PCO pilot scale systems . . . . . . . . . . . . . . . . . . . . . . . . . . . Air stripper application . . . . . . . . . . . . . . . . . . . . . . . . . . . . Paint booth application . . . . . . . . . . . . . . . . . . . . . . . . . . . . Field test results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Air stripper application . . . . . . . . . . . . . . . . . . . . . . . . . . . . Paint booth application . . . . . . . . . . . . . . . . . . . . . . . . . . . . Comparison with other treatment systems . . . . . . . . . . . . . . . . . . CONTAMINATION CLEANUP: NATURAL ATTENUATION AND ADVANCED REMEDIATION TECHNOLOGIES . . . . . . . . . . . Natural attenuation of chlorinated solvents in ground water . . . . . . . . .

. . . . . . . .

. . . . . . . .

1559 1560 1560 1562 1564 1564 1566 1568

. . 1571 . . 1571

HANADI S. RIFAI, CHARLES J. NEWELL, TODD H. WIEDEMEIER

23.1.1 23.1.2 23.1.2.1 23.1.2.2 23.1.2.3 23.1.2.4 23.1.2.5 23.1.2.6 23.1.2.7 23.1.2.8 23.1.3 23.1.3.1 23.1.3.1.1 23.1.3.1.2 23.1.3.2 23.1.3.2.1 23.1.3.2.2 23.1.3.2.3 23.1.4 23.1.4.1 23.1.4.2 23.1.4.3 23.1.5 23.1.5.1 23.1.5.2 23.1.5.2.1 23.1.5.2.2 23.1.5.2.3 23.1.5.2.4 23.1.6 23.1.6.1 23.1.6.2 23.1.6.2.1 23.1.6.3 23.1.6.4 23.2

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Natural attenuation processes affecting chlorinated solvent plumes . . . . . . Advection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sorption. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . One-dimensional advection-dispersion equation with retardation . . . . . . . Dilution (recharge) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Volatilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrolysis and dehydrohalogenation . . . . . . . . . . . . . . . . . . . . . Reduction reactions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biodegradation of chlorinated solvents . . . . . . . . . . . . . . . . . . . . Halorespiration or reductive dechlorination using hydrogen. . . . . . . . . . Stoichiometry of reductive dechlorination . . . . . . . . . . . . . . . . . . . Chlorinated solvents that are amenable to halorespiration . . . . . . . . . . . Oxidation of chlorinated solvents . . . . . . . . . . . . . . . . . . . . . . . Direct aerobic oxidation of chlorinated compounds . . . . . . . . . . . . . . Aerobic cometabolism of chlorinated compounds . . . . . . . . . . . . . . . Anaerobic oxidation of chlorinated compounds . . . . . . . . . . . . . . . . Biodegradation rates for chlorinated solvents . . . . . . . . . . . . . . . . . Michaelis-Menten rates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zero-order rates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . First-order rate constants . . . . . . . . . . . . . . . . . . . . . . . . . . . . Geochemical evidence of natural bioremediation at chlorinated solvent sites . Assessing reductive dechlorination at field sites . . . . . . . . . . . . . . . . Plume classification schemes. . . . . . . . . . . . . . . . . . . . . . . . . . Type 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Type 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Type 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mixed environments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chlorinated solvent plumes - case studies of natural attenuation . . . . . . . Plume databases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Modeling chlorinated solvent plumes . . . . . . . . . . . . . . . . . . . . . BIOCHLOR natural attenuation model . . . . . . . . . . . . . . . . . . . . RT3D numerical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . CS case study - The Plattsburgh Air Force Base . . . . . . . . . . . . . . . . Remediation technologies and approaches for managing sites impacted by hydrocarbons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1571 1572 1572 1573 1574 1577 1577 1578 1579 1581 1581 1582 1585 1585 1586 1586 1587 1587 1588 1588 1590 1591 1599 1599 1599 1599 1600 1601 1601 1602 1602 1605 1605 1609 1611

. 1617

BARRY J. SPARGO, JAMES G. MUELLER

23.2.1 23.2.1.1 23.2.1.2

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1617 Understanding HC and CHC in the environment . . . . . . . . . . . . . . . . 1617 Sources of HC in the environment . . . . . . . . . . . . . . . . . . . . . . . . 1617

Table of contents

23.2.1.3 23.2.2 23.2.2.1 23.2.2.1.1 23.2.2.2 23.2.2.2.1

xxv

23.2.4 24

Sources of CHC in the environment . . . . . . . . . . . . . . . . . . . . . . In situ biotreatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microbial-enhanced natural attenuation/bioremediation . . . . . . . . . . . . Case study - Cooper River Watershed, Charleston, SC, USA . . . . . . . . . Phytoremediation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Case study - phytoremediation for CHCs in groundwater at a chemical plant in Louisiana. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . In situ treatment technologies . . . . . . . . . . . . . . . . . . . . . . . . . Product recovery via GCW technology . . . . . . . . . . . . . . . . . . . . Case study - GCW recovery of creosote, Cabot/Kopper’s Superfund Site, Gainesville, FL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Surfactant enhanced product recovery . . . . . . . . . . . . . . . . . . . . . Case study - Surfactant-aided chlorinated HC DNAPL recovery, Hill Air Force Base, Ogden, Utah . . . . . . . . . . . . . . . . . . . . . . . Foam-enhanced product recovery . . . . . . . . . . . . . . . . . . . . . . . Thermal desorption - Six Phase Heating . . . . . . . . . . . . . . . . . . . . Case study - Six-Phase Heating removal of CHC at a manufacturing facility near Chicago, IL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . In situ steam enhanced extraction (Dynamic Underground Stripping) . . . . In situ permeable reactive barriers (funnel and gate). . . . . . . . . . . . . . Case study - CHC remediation using an in situ permeable reactive barrier at Naval Air Station Moffett Field, CA . . . . . . . . . . . . . . . . . . . . Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . PROTECTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

24.1 24.2 24.3 25

Gloves . . . . . . . . . . . . . . . . . . . . . . . . . . . Suit materials . . . . . . . . . . . . . . . . . . . . . . . . Respiratory protection . . . . . . . . . . . . . . . . . . . NEW TRENDS BASED ON PATENT LITERATURE

25.1 25.2 25.3 25.4 25.5 25.6 25.7 25.8 25.9 25.10 25.11 25.12 25.13 25.14 25.15 25.16 25.17

New solvents . . . . . . . . . Adhesives. . . . . . . . . . . Aerospace. . . . . . . . . . . Agriculture . . . . . . . . . . Asphalt . . . . . . . . . . . . Automotive applications . . . Coil coating. . . . . . . . . . Composites and laminates . . Cosmetics. . . . . . . . . . . Cleaning . . . . . . . . . . . Fibers . . . . . . . . . . . . . Furniture and wood coatings . Paper . . . . . . . . . . . . . Printing . . . . . . . . . . . . Stone and concrete . . . . . . Wax . . . . . . . . . . . . . . Summary . . . . . . . . . . . ACKNOWLEDGMENTS . INDEX . . . . . . . . . . . .

23.2.3 23.2.3.1 23.2.3.1.1 23.2.3.2 23.2.3.2.1 23.2.3.3 23.2.3.4 23.2.3.4.1 23.2.3.5 23.2.3.6 23.2.3.6.1

. . . . .

1618 1618 1618 1620 1622

. 1622 . 1623 . 1623 . 1624 . 1625 . 1625 . 1626 . 1626 . 1627 . 1628 . 1628 . 1628 . 1629 . 1631

GEORGE WYPYCH

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

1631 1633 1633 1637

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

1637 1638 1640 1640 1640 1641 1641 1642 1643 1644 1645 1646 1647 1647 1648 1648 1649 1653 1657

GEORGE WYPYCH

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . .

Preface Although the chemical industry can trace its roots into antiquity, it was during the industrial revolution that it started to become an actual industry and began to use the increased knowledge of chemistry as a science and technology to produce products that were needed by companion industries and consumers. These commercial efforts resulted in the synthesis of many new chemicals. Quite quickly, in these early days, previously unknown materials or materials that had been present only in low concentrations, were now in contact with people in highly concentrated forms and in large quantities. The people had little or no knowledge of the effects of these materials on their bodies and the natural biological and physical processes in the rivers and oceans, the atmosphere, and in the ground. Until the end of the nineteenth century these problems were not addressed by the chemical industry and it is only recently that the industry began to respond to public criticism and political efforts. Legal restrictions aimed at preserving the quality of life have been directed at health, safety and longevity issues and the environment. Solvents have always been mainstays of the chemical industry and because of their widespread use and their high volume of production they have been specifically targeted by legislators throughout the world. The restrictions range from total prohibition of production and use, to limits placed on vapor concentrations in the air. As with any arbitrary measures some solvents have been damned unfairly. However, there is no question that it is best to err on the side of safety if the risks are not fully understood. It is also true that solvents should be differentiated based on their individual properties. This book is intended to provide a better understanding of the principles involved in solvent selection and use. It strives to provide information that will help to identify the risks and benefits associated with specific solvents and classes of solvents. The book is intended to help the formulator select the ideal solvent, the safety coordinator to safeguard his or her coworkers, the legislator to impose appropriate and technically correct restrictions and the student to appreciate the amazing variety of properties, applications and risks associated with the more than one thousand solvents that are available today. By their very nature, handbooks are intended to provide exhaustive information on the subject. While we agree that this is the goal here, we have attempted to temper the impact of information, which may be too narrow to make decision. Many excellent books on solvents have been published in the past and most of these are referenced in this book. But of all these books none has given a comprehensive overview of all aspects of solvent use. Access to comprehensive data is an essential part of solvent evaluation and it has been a hallmark of such books to provide tables filled with data to the point at which 50 to 95% of the book is data. This approach seems to neglect a fundamental requirement of a handbook - to provide the background, explanations and clarifications that are needed to convert data to information and assist the reader in gaining the knowledge to make a decision on selecting a process or a solvent. Unfortunately, to meet the goal of providing both the data and the fundamental explanations that are needed, a book of 4,000 to 5,000 pages might be required. Even if this was possible, much of the data would fall out of date quite quickly. For example, a factor that defines solvent safety such as threshold limit

xxviii

Preface

values (TLVs) for worker exposure or some single toxicity determinants may change frequently. This book would be huge and it would have to be updated frequently to continue to claim that it is current. What we have attempted to do here is to give you a book with a comprehensive and extensive analysis of all current information on solvents then use other media to present the supporting data on individual solvents. These data are provided on a CD-ROM as a searchable database. Data are provided on more than 1140 solvents in 110 fields of data. The medium permits frequent updates. If the same data were presented in book form, more than 2,000 pages would be needed which exceeds the size of any data in handbook form offered to date. The best approach in presenting an authoritative text for such a book is to have it written by experts in their fields. This book attracted well-known experts who have written jointly 47 books and authored or coauthored hundreds of papers on their areas of expertise. The authors have made their contributions to this book in late 1999 and early 2000 providing the most current picture of the technology. Their extreme familiarity with their subjects enables them to present information in depth and detail, which is essential to the reader’s full understanding of the subject. The authors were aware of the diversity of potential readers at the outset and one of their objectives was to provide information to various disciplines expressed in a way that all would understand and which would deal with all aspects of solvent applications. We expect professionals and students from a wide range of businesses, all levels of governments and academe to be interested readers. The list includes solvent manufacturers, formulators of solvent containing products, industrial engineers, analytical chemists, government legislators and their staffs, medical professionals involved in assessing the impact on health of solvents, biologists who are evaluating the interactions of solvents with soil and water, environmental engineers, industrial hygienists who are determining protective measures against solvent exposure, civil engineers who design waste disposal sites and remediation measures, people in industries where there are processes which use solvents and require their recovery and, perhaps most important, because understanding brings improvements, those who teach and learn in our universities, colleges and schools. A growing spirit of cooperation is evident between these groups and this can be fostered by providing avenues of understanding based on sharing data and information on common problems. We hope to provide one such avenue with this book. We have tried to present a balanced picture of solvent performance by dealing not only with product performance and ease of processing but also by giving environmental and health issues full consideration. Data and information on known products and processes should be cornerstones of the understanding of a technology but there is another aspect of technology, which can lead to advances and improvements in utility, safety and in safeguarding the environment. This must come from you, the reader. It is your ideas and creative thinking that will bring these improvements. The authors have crammed their ideas into the book and we hope these will stimulate responsible and effective applications of solvents. Francis Bacon wrote, “The end of our foundation is the knowledge of causes, and the secret motion of things, and the enlarging of the bound of human Empire, to the effecting of all things possible.” Today there are few technical activities that do not employ solvents. Almost all industries, almost all consumer products, almost everything we use can, if analyzed, be shown to

Preface

xxix

contain or to have used in its processing, a solvent. Solvent elimination need never be a technical objective. Rather, we need to use our increasing understanding and knowledge to find the safest and the most effective means of meeting our goals. I would like to thank the authors for their relentless efforts to explain the difficult in an interesting way. In advance, I would like to thank the reader for choosing this book and encourage her or him to apply the knowledge to make our world a better, more livable place. George Wypych Toronto, August 3, 2000

1

Introduction Christian Reichardt Department of Chemistry, Philipps University, Marburg, Germany

Chemical transformations can be performed in a gas, liquid, or solid phase, but, with good reasons, the majority of such reactions is carried out in the liquid phase in solution. At the macroscopic level, a liquid is the ideal medium to transport heat to and from exo- and endothermic reactions. From the molecular-microscopic point of view, solvents break the crystal lattice of solid reactants, dissolve gaseous or liquid reactants, and they may exert a considerable influence over reaction rates and the positions of chemical equilibria. Because of nonspecific and specific intermolecular forces acting between the ions or molecules of dissolved reactants, activated complexes as well as products and solvent molecules (leading to differential solvation of all solutes), the rates, equilibria, and the selectivity of chemical reactions can be strongly influenced by the solvent. Other than the fact that the liquid medium should dissolve the reactants and should be easily separated from the reaction products afterwards, the solvent can have a decisive influence on the outcome (i.e., yield and product distribution) of the chemical reaction under study. Therefore, whenever a chemist wishes to perform a certain chemical reaction, she/he has to take into account not only suitable reaction partners and their concentrations, the proper reaction vessel, the appropriate reaction temperature, and, if necessary, the selection of the right reaction catalyst but also, if the planned reaction is to be successful, the selection of an appropriate solvent or solvent mixture. Solvent effects on chemical reactivity have been studied for more than a century, beginning with the pioneering work of Berthelot and Saint Gilles1 in Paris in 1862 on esterification reactions and with that of Menschutkin2 in St. Petersburg in 1880 on the quaternization of tertiary amines by haloalkanes. At this time Menschutkin remarked that “a reaction cannot be separated from the medium in which it is performed... Experience shows that solvents exert considerable influence on reaction rates.” Today, we can suggest a striking example to reinforce his remark, the rate of the unimolecular heterolysis of 2-chloro-2-methylpropane observed in water and benzene increases by a factor of approximately1011 when the nonpolar benzene is replaced by water.3,4 The influence of solvents on the position of chemical equilibria was discovered in 1896 by Claisen5 in Aachen, Knorr6 in Jena, Wislicenus7 in Wòrzburg, and Hantzsch8 in Wòrzburg. They investigated almost simultaneous but independent of one another the keto-enol tautomerism of 1,3-dicarbonyl compounds and the nitro-isonitro tautomerism of primary and secondary

2

Christian Reichardt

aliphatic nitro compounds. With this example, the enol content of acetylacetone increases from 62 to 95 % when acetonitrile is substituted with n-hexane.3,9 The proper solvent and solvent mixture selection is not only important for chemical but also for physical processes such as recrystallization, all extraction processes, partitioning, chromatographic separations, phase-transfer catalytic reactions, etc. Of particular interest in this context is the influence of solvents on all types of light absorption processes, e.g., on UV/Vis, IR, ESR, and NMR spectra, caused by differential solvation of the ground and excited states of the absorbing species.3,12 In 1878, Kundt10 in Zòrich proposed the rule that increasing dispersion interactions between the absorbing solute and the solvent lead in general to a bathochromic shift of an UV/Vis absorption band. Later, in 1922, Hantzsch11 termed the solvent-dependence of UV/Vis absorption spectra “solvatochromism”. UV/Vis absorption of solute molecules can be influenced not only by the surrounding solvent sphere, but also by other entities in the surroundings such as solids, polymers, glasses, and surfaces. In order to emphasize this influence, the use of the more general term “perichromism” (from Greek peri = around) has been recommended.12,13 A typical, more recent, example of extraordinary solvatochromism is the intramolecular charge-transfer Vis-absorption of 2,6-diphenyl-4-(2,4,6-triphenyl-l-pyridinio)phenolate, a zwitterionic betaine dye: its corresponding absorption band is shifted from λmax = 810 nm to λmax = 453 nm (∆λ = 357 nm) when diphenyl ether is replaced by water as solvent.3,12 Such solvatochromic dyes can be used as empirical solvent polarity indicators.12 The number of solvents generally available to chemists working in research and industrial laboratories is between 250 and 3003,14 (there is an infinite number of solvent mixtures), and this number is increasing. More recently and for obvious reasons, the search for new solvents has been intensified: peroxide-forming solvents are being substituted by solvents which are more stable against oxidation (e.g., diethyl ether by t-butyl methyl ether or by formaldehyde dialkyl acetals), toxic solvents are being replaced by nontoxic ones (e.g., the cancerogenic hexamethylphosphoric triamide, HMPT, by N,N'-dimethylpropyleneurea, DMPU15) and environmentally dangerous solvents by benign ones (e.g., tetrachloromethane by perfluorohexane16). The development of modern solvents for organic syntheses is the subject of much current research.17 Amongst these modern solvents, also called “neoteric solvents” (neoteric = recent, new, modern) in contrast to the classical ones, are ionic liquids (i.e., room-temperature liquid salts such as 1-ethyl-3-methylimidazolium tetra-chloroaluminates18,19), supercritical-fluid solvents, SCF, (such as SCF carbon dioxide20,21), and perfluorinated solvents (e.g., partially or perfluorinated hydrocarbons as used in so-called “fluorous biphase catalysis reactions”, making possible mono-phase reactions and a two-phase separation of catalyst and reaction products22-24). Even plain water has found a magnificent renaissance as a solvent for organic reactions.25,26 These efforts have also recently strengthened the search for completely solvent-free reactions, thus avoiding the use of expensive, toxic, and environmentally problematic solvents.27,28 With respect to the large and still increasing number of valuable solvents useful for organic syntheses, a chemist needs, in addition to his experience and intuition, to have general rules, objective criteria, and the latest information about the solvents' physical, chemical, and toxicological properties for the selection of the proper solvent or solvent mixture for a planned reaction or a technological process. To make this often cumbersome and time-consuming task easier, this “Handbook of Solvents” with its twenty-five chapters is designed to provide a comprehensive source of information on solvents over a broad range

1 Introduction

3

of applications. It is directed not only to chemists working in research laboratories, but also to all industries using solvents for various purposes. A particular advantage is that the printed handbook is accompanied by a compact-disc (CD-ROM) containing additional solvent databases with hundred ten fields for over eleven hundred solvents. This makes large data sets easily available for quick search and retrieval and frees the book text from bulky tables, thus giving more room for a thorough description of the underlying theoretical and practical fundamental subjects. Fundamental principles governing the use of solvents (i.e., chemical structure, molecular design as well as physical and chemical properties of solvents) are given in Chapter 2. Solvent classification, methods of solvent manufacture together with properties and typical applications of various solvents are provided in Chapter 3. Chapters 4, 5 and 6 deal with all aspects of the dissolution of materials in solvents as well as with the solubility of selected systems (e.g., polymers and elastomers) and the influence of the solute's molecular structure on its solubility behavior. In particular, the valuable solubility-parameter concept is extensively treated in these chapters. All aspects of solvent transport within polymeric system and the drying of such polymeric systems, including coated films, are described in Chapter 7. The fundamentals of the interaction forces acting between ions or molecules of the solvents themselves and between solutes and solvents in solutions are presented in Chapter 8. Chapter 9 deals with the corresponding properties of solvent mixtures. Specific solute/solvent interactions, particularly Lewis acid/base interactions between electron-pair donors (EPD) and electron-pair acceptors (EPA), are reviewed in Chapter 10, together with the development of empirical scales of solvent polarity and Lewis acidity/basicity, based on suitable solvent-dependent reference processes, and their application for the treatment of solvent effects. The theory for solvent effects on electronic properties is provided in Chapter 11 and extended to solvent-dependent properties of solutes such as fluorescence spectra, ORD and CD spectra. Aggregation, swelling of polymers, their conformations, the viscosity of solutions and other solvent-related properties are treated in Chapter 12. A review concerning solvent effects on various types of chemical reactivity is given in Chapter 13, along with a discussion of the effects of solvent on free-radical polymerization and phase-transfer catalysis reactions. The second part of this handbook (Chapters 14-25) is devoted more to the industrial use of solvents. Formulating with solvents applied in a broad range of industrial areas such as biotechnology, dry cleaning, electronic industry, food industry, paints and coatings, petroleum refining industry, pharmaceutical industry, textile industry, to mention only a few, is extensively described in Chapter 14. Standard and special methods of solvent detection and solvent analysis as well as the problem of residual solvents in various products, particularly in pharmaceutical ones, are the topics of Chapters 15 and 16. At present, large-scale chemical manufacturing is facing serious solvent problems with respect to environmental concerns. National and international regulations for the proper use of hazardous solvents are becoming increasingly stringent and this requires the use of environmentally more benign but nevertheless economical liquid reaction media. This has enormously stimulated the search for such new solvent systems within the framework of so-called green chemistry. Supercritical fluids, SCF,20,21 and ionic liquids (room temperature liquid salts)18,19 have been known and have been the subject of scientific interest for a long time. It is only recently, however, that the potential benefits of these materials in solvent applications have been realized.17 This handbook includes in Chapters 17-25 all

4

Christian Reichardt

the knowledge necessary for a safe handling of solvents in research laboratories and in large-scale manufacturing, beginning with the environmental impact of solvents on water, soil, and air in Chapter 17, followed by considerations about safe solvent concentrations and the risks of solvent exposure in various industrial environments in Chapter 18. Chapter 19 summarizes the corresponding legal regulations, valid for North America and Europe, and Chapter 20 describes in detail the toxic effects of solvent exposure to human beings. Authors specializing in different fields of solvent toxicity give the most current information on the effect of solvent exposure from the point of view of neurotoxicity, reproductive and maternal effects, nephrotoxicity, cancerogenicity, hepatotoxicity, chromosomal aberrations, and toxicity to brain, lungs, and heart. This information brings both the results of documented studies and an evaluation of risk in different industrial environments in a comprehensive but easy to understand form to engineers and decision-makers in industry. Chapter 21 is focused on the substitution of harmful solvents by safer ones and on the development of corresponding new technological processes. Chapter 22 describes modern methods of solvent recovery, solvent recycling. When recycling is not possible, then solvents have to be destroyed by incineration or other methods of oxidation, as outlined in Chapter 22. Chapter 23 describes natural attenuation of solvents in groundwater and advanced remediation technologies as well as management strategies for sites impacted by solvent contamination. Protection from contact with solvents and their vapors is discussed in Chapter 24. Finally, new trends in solvent chemistry and applications based on the recent patent literature are discussed in Chapter 25. In most cases, the intelligent choice of the proper solvent or solvent mixture is essential for the realization of certain chemical transformations and physical processes. This handbook tries to cover all theoretical and practical information necessary for this often difficult task for both academic and industrial applications. It should be used not only by chemists, but also by physicists, chemical engineers, and technologists as well as environmental scientists in academic and industrial institutions. It is to be hoped that the present compilation of all relevant aspects connected with the use of solvents will also stimulate further basic and applied research in the still topical field of the physics and chemistry of liquid media.

REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

M Berthelot, L P¾an de Saint Gilles, Ann. Chim. Phys., 3. S¾r., 65, 385 (1862); ibid. 66, 5 (1862); ibid. 68, 255 (1863). N Menschutkin, Z. Phys. Chem., 5, 589 (1890); ibid. 6, 41 (1890); ibid. 34, 157 (1900). C Reichardt, Solvents and Solvent Effects in Organic Chemistry, 2nd ed., VCH, Weinheim, 1988. (a) G F Dvorko, E A Ponomareva, Usp. Khim., 53, 948 (1984); Russ. Chem. Rev., 53, 547 (1984); (b) M H Abraham, Pure Appl. Chem., 57, 1055 (1985); and references cited therein. L Claisen, Justus Liebigs Ann. Chem., 291, 25 (1896). L Knorr, Justus Liebigs Ann. Chem., 293, 70 (1896). W Wislicenus, Justus Liebigs Ann. Chem., 291, 147 (1896). A Hantzsch, O W Schultze, Ber. Dtsch. Chem. Ges., 29, 2251 (1896). M T Rogers, J L Burdett, Can. J Chem., 43, 1516 (1965). A Kundt, Poggendorfs Ann. Phys. Chem. N. F., 4, 34 (1878); Chem. Zentralbl., 498 (1878). A Hantzsch, Ber. Dtsch. Chem. Ges., 55, 953 (1922). C Reichardt, Chem. Rev., 94, 2319 (1994). Prof. E M Kosower, Tel Aviv, private communication to C.R. Y Marcus, The Properties of Solvents, Wiley, Chichester, 1998. (a) Editorial, Chimia, 39, 147 (1985); (b) D Seebach, Chemistry in Britain, 21, 632 (1985). S M Pereira, G P Sauvage, G. W. Simpson, Synth. Commun., 25, 1023 (1995).

Introduction

17 18 19

20 21 22 23 24 25 26 27 28

5

P Knochel (Ed.), Modern Solvents in Organic Synthesis, Topics in Current Chemistry, Vol. 206, Springer, Berlin, 1999. Y Chauvin, H Olivier-Bourbigou, CHEMTECH, 25(9), 26 (1995). (a) K R Seddon, Kinetika i Kataliz, 37, 743 (1996); Kinetics and Catalysis, 37, 693 (1996); Chem. Abstr., 125, 285927s (1996); (b) K R Seddon, J. Chem. Technol. Biotechnol., 68, 351 (1997); Chem. Abstr., 126, 306898w (1997). R Noyori (Ed.), Supercritical Fluids, Chem. Rev., 99, 353-633 (1999). W Leitner, Top. Curr. Chem., 206, 107 (1999). B Cornils, Angew. Chem., 109, 2147 (1997); Angew. Chem., Int. Ed. Engl., 36, 2057 (1997). B Betzemeier, P Knochel, Top. Curr. Chem., 206, 61 (1999). J J Maul, P J Ostrowski, G A Ublacker, B Linclau, D P Curran, Top. Curr. Chem., 206, 79 (1999). P A Grieco, Organic Synthesis in Water, Blackie Academic and Professional, Hampshire, 1998. A Lubineau and J. Aug¾, Top. Curr. Chem., 206, 1 (1999). J O Metzger, Angew, Chem., 110, 3145 (1998); Angew. Chem., Int. Ed. Engl., 37, 2975 (1998). A Loupy, Top. Curr. Chem., 206, 153 (1999).

2

Fundamental Principles Governing Solvents Use 2.1 SOLVENT EFFECTS ON CHEMICAL SYSTEMS Estanislao Silla, Arturo Arnau and Iñaki TuñóN Department of Physical Chemistry, University of Valencia, Burjassot (Valencia), Spain

2.1.1 HISTORICAL OUTLINE According to a story, a little fish asked a big fish about the ocean, since he had heard it being talked about but did not know where it was. Whilst the little fish’s eyes turned bright and shiny full of surprise, the old fish told him that all that surrounded him was the ocean. This story illustrates in an eloquent way how difficult it is to get away from every day life, something of which the chemistry of solvents is not unaware. The chemistry of living beings and that which we practice in laboratories and factories is generally a chemistry in solution, a solution which is generally aqueous. A daily routine such as this explains the difficulty which, throughout the history of chemistry, has been encountered in getting to know the effects of the solvent in chemical transformations, something which was not achieved in a precise way until well into the XX century. It was necessary to wait for the development of experimental techniques in vacuo to be able to separate the solvent and to compare the chemical processes in the presence and in the absence of this, with the purpose of getting to know its role in the chemical transformations which occur in its midst. But we ought to start from the beginning. Although essential for the later cultural development, Greek philosophy was basically a work of the imagination, removed from experimentation, and something more than meditation is needed to reach an approach on what happens in a process of dissolution. However, in those remote times, any chemically active liquid was included under the name of “divine water”, bearing in mind that the term “water” was used to refer to anything liquid or dissolved.1 Parallel with the fanciful search for the philosopher’s stone, the alchemists toiled away on another impossible search, that of a universal solvent which some called “alkahest” and others referred to as “menstruum universale”, which term was used by the very Paracelsus (1493-1541), which gives an idea of the importance given to solvents during that dark and obscurantist period. Even though the “menstruum universale” proved just as elusive as the philosopher’s stone, all the work carried out by the alchemists in search of these

8

Estanislao Silla, Arturo Arnau and Iñaki Tuñón

illusionary materials opened the way to improving the work in the laboratory, the development of new methods, the discovery of compounds and the utilization of novel solvents. One of the tangible results of all that alchemistry work was the discovery of one of the first experimental rules of chemistry: “similia similibus solvuntor”, which reminds us of the compatibility in solution of those substances of similar nature. Even so, the alchemistry only touched lightly on the subject of the role played by the solvent, with so many conceptual gulfs in those pre-scientific times in which the terms dissolution and solution referred to any process which led to a liquid product, without making any distinction between the fusion of a substance - such as the transformation of ice into liquid water -, mere physical dissolution - such as that of a sweetener in water - or the dissolution which takes place with a chemical transformation - such as could be the dissolution of a metal in an acid. This misdirected vision of the dissolution process led the alchemists down equally erroneous collateral paths which were prolonged in time. Some examples are worth quoting: Hermann Boerhaave (1688-1738) thought that dissolution and chemical reaction constituted the same reality; the solvent, (menstruum), habitually a liquid, he considered to be formed by diminutive particles moving around amongst those of the solute, leaving the interactions of these particles dependent on the mutual affinities of both substances.2 This paved the way for Boerhaave to introduce the term affinity in a such a way as was conserved throughout the whole of the following century.3 This approach also enabled Boerhaave to conclude that combustion was accompanied by an increase of weight due to the capturing of “particles” of fire, which he considered to be provided with weight by the substance which was burned. This explanation, supported by the well known Boyle, eased the way to considering that fire, heat and light were material substances until when, in the XIX century, the modern concept of energy put things in their place.4 Even Bertollet (1748-1822) saw no difference between a dissolution and a chemical reaction, which prevented him from reaching the law of definite proportions. It was Proust, an experimenter who was more exacting and capable of differentiating between chemical reaction and dissolution, who made his opinion prevail:

“The dissolution of ammonia in water is not the same as that of hydrogen in azote (nitrogen), which gives rise to ammonia”5 There were also alchemists who defended the idea that the substances lost their nature when dissolved. Van Helmont (1577-1644) was one of the first to oppose this mistaken idea by defending that the substance dissolved remains in the solution in aqueous form, it being possible to recover it later. Later, the theories of osmotic pressure of van´t Hoff (1852-1911) and that of electrolytic dissociation of Arrhenius (1859-1927) took this approach even further. Until almost the end of the XIX century the effects of the solvent on the different chemical processes did not become the object of systematic study by the experimenters. The effect of the solvent was assumed, without reaching the point of awakening the interest of the chemists. However, some chemists of the XIX century were soon capable of unraveling the role played by some solvents by carrying out experiments on different solvents, classified according to their physical properties; in this way the influence of the solvent both on chemical equilibrium and on the rate of reaction was brought to light. Thus, in 1862, Berthelot and Saint-Gilles, in their studies on the esterification of acetic acid with ethanol,

2.1 Solvent effects on chemical systems

9

discovered that some solvents, which do not participate in the chemical reaction, are capable of slowing down the process.6 In 1890, Menschutkin, in a now classical study on the reaction of the trialkylamines with haloalcans in 23 solvents, made it clear how the choice of one or the other could substantially affect the reaction rate.7 It was also Menschutkin who discovered that, in reactions between liquids, one of the reactants could constitute a solvent inadvisable for that reaction. Thus, in the reaction between aniline and acetic acid to produce acetanilide, it is better to use an excess of acetic acid than an excess of aniline, since the latter is a solvent which is not very favorable to this reaction. The fruits of these experiments with series of solvents were the first rules regarding the participation of the solvent, such as those discovered by Hughes and Ingold for the rate of the nucleophilic reactions.8 Utilizing a simple electrostatic model of the solute - solvent interactions, Hughes and Ingold concluded that the state of transition is more polar than the initial state, that an increase of the polarity of the solvent will stabilize the state of transition with respect to the initial state, thus leading to an increase in the reaction rate. If, on the contrary, the state of transition is less polar, then the increase of the polarity of the solvent will lead to a decrease of the velocity of the process. The rules of Hughes-Ingold for the nucleophilic aliphatic reactions are summarized in Table 2.1.1. Table 2.1.1. Rules of Hughes-Ingold on the effect of the increase of the polarity of the solvent on the rate of nucleophilic aliphatic reactions Mechanism

Initial state -

S N2

-

Effect on the reaction rate

Y + RX

[Y--R--X]

slight decrease

Y + RX

[Y--R--X]

large increase

-

+

Y + RX

+

S N1

State of transition

[Y--R--X] +

large decrease

Y + RX

[Y--R--X]

slight decrease

RX

[R--X]

large increase

+

RX

[R--X]

+

slight decrease

In 1896 the first results about the role of the solvent on chemical equilibria were obtained, coinciding with the discovery of the keto-enolic tautomerism.9 Claisen identified the medium as one of the factors which, together with the temperature and the substituents, proved to be decisive in this equilibrium. Soon systematic studies began to be done on the effect of the solvent in the tautomeric equilibria. Wislicenus studied the keto-enolic equilibrium of ethylformylphenylacetate in eight solvents, concluding that the final proportion between the keto form and the enol form depended on the polarity of the solvent.10 This effect of the solvent also revealed itself in other types of equilibria: acid-base, conformational, those of isomerization and of electronic transfer. The acid-base equilibrium is of particular interest. The relative scales of basicity and acidity of different organic compounds and homologous families were established on the basis of measurements carried out in solution, fundamentally aqueous. These scales permitted establishing hypotheses regarding the effect of the substituents on the acidic and basic centers, but without being capable of separating this from the effect of the solvent. Thus, the scale obtained in solution for the acidity of

10


the α-substituted methyl alcohols [(CH3)3COH > (CH3)2CHOH > CH3CH2OH > CH3OH]11 came into conflict with the conclusions extracted from the measurements of movements by NMR.12 The irregular order in the basicity of the methyl amines in aqueous solution also proved to be confusing [NH3 < CH3NH2 < (CH3)2NH > (CH3)3N],13 since it did not match any of the existing models on the effects of the substituents. These conflicts were only resolved when the scales of acidity-basicity were established in the gas phase. On carrying out the abstraction of the solvent an exact understanding began to be had of the real role it played. The great technological development which arrived with the XIX century has brought us a set of techniques capable of giving accurate values in the study of chemical processes in the gas phase. The methods most widely used for these studies are: • The High Pressure Mass Spectrometry, which uses a beam of electron pulses14 • The Ion Cyclotron Resonance and its corresponding Fourier Transform (FT-ICR)15 • The Chemical Ionization Mass Spectrometry, in which the analysis is made of the kinetic energy of the ions, after generating them by collisions16 • The techniques of Flowing Afterglow, where the flow of gases is submitted to ionization by electron bombardment17-19 All of these techniques give absolute values with an accuracy of ±(2-4) Kcal/mol and of ±0.2 Kcal/mol for the relative values.20 During the last decades of XX century the importance has also been made clear of the effects of the solvent in the behavior of the biomacromolecules. To give an example, the influence of the solvent over the proteins is made evident not only by its effects on the structure and the thermodynamics, but also on the dynamics of these, both at local as well as at global level.21 In the same way, the effect of the medium proves to be indispensable in explaining a large variety of biological processes, such us the rate of interchange of oxygen in myoglobin.22 Therefore, the actual state of development of chemistry, as much in its experimental aspect as in its theoretical one, allows us to identify and analyze the influence of the solvent on processes increasingly more complex, leaving the subject open for new challenges and investigating the scientific necessity of creating models with which to interpret such a wide range of phenomena as this. The little fish became aware of the ocean and began explorations. 2.1.2 CLASSIFICATION OF SOLUTE-SOLVENT INTERACTIONS Fixing the limits of the different interactions between the solute and the solvent which envelopes it is not a trivial task. In the first place, the liquid state, which is predominant in the majority of the solutions in use, is more difficult to comprehend than the solid state (which has its constitutive particles, atoms, molecules or ions, in fixed positions) or the gaseous state (in which the interactions between the constitutive particles are not so intense). Moreover, the solute-solvent interactions, which, as has already been pointed out, generally happen in the liquid phase, are half way between the predominant interactions in the solid phase and those which happen in the gas phase, too weak to be likened with the physics of the solid state but too strong to fit in with the kinetic theory of gases. In the second place, dissecting the solute-solvent interaction into different sub-interactions only serves to give us an approximate idea of the reality and we should not forget that, in the solute-solvent interaction, the all is not the sum of the parts. In the third place, the world of the solvents is very varied from those which have a very severe internal structure, as in the case of water, to those


11

whose molecules interact superficially, as in the case of the hydrocarbons. At all events, there is no alternative to meeting the challenge face to face. If we mix a solute and a solvent, both being constituted by chemically saturated molecules, their molecules attract one another as they approach one another. This interaction can only be electrical in its nature, given that other known interactions are much more intense and of much shorter range of action (such as those which can be explained by means of nuclear forces) or much lighter and of longer range of action (such as the gravitational force). These intermolecular forces usually also receive the name of van der Waals forces, from the fact that it was this Dutch physicist, Johannes D. van der Waals (1837-1923), who recognized them as being the cause of the non-perfect behavior of the real gases, in a period in which the modern concept of the molecule still had to be consolidated. The intermolecular forces not only permit the interactions between solutes and solvents to be explained but also determine the properties of gases, liquids and solids; they are essential in the chemical transformations and are responsible for organizing the structure of biological molecules. The analysis of solute-solvent interactions is usually based on the following partition scheme: ∆E = ∆E i + ∆E ij + ∆E jj

[2.1.1]

where i stands for the solute and j for the solvent.This approach can be maintained while the identities of the solute and solvent molecules are preserved. In some special cases (see below in specific interactions) it will be necessary to include some solvent molecules in the solute definition. The first term in the above expression is the energy change of the solute due to the electronic and nuclear distortion induced by the solvent molecule and is usually given the name solute polarization. ∆Eij is the interaction energy between the solute and solvent molecules. The last term is the energy difference between the solvent after and before the introduction of the solute. This term reflects the changes induced by the solute on the solvent structure. It is usually called cavitation energy in the framework of continuum solvent models and hydrophobic interaction when analyzing the solvation of nonpolar molecules. The calculation of the three energy terms needs analytical expressions for the different energy contributions but also requires knowledge of solvent molecules distribution around the solute which in turn depends on the balance between the potential and the kinetic energy of the molecules. This distribution can be obtained from diffraction experiments or more usually is calculated by means of different solvent modelling. In this section we will comment on the expression for evaluating the energy contributions. The first two terms in equation [2.1.1] can be considered together by means of the following energy partition : ∆E i + ∆E ij = ∆E el + ∆E pol + ∆E d − r

[2.1.2]

Analytical expressions for the three terms (electrostatic, polarization and dispersion-repulsion energies) are obtained from the intermolecular interactions theory. 2.1.2.1 Electrostatic The electrostatic contribution arises from the interaction of the unpolarized charge distribution of the molecules. This interaction can be analyzed using a multipolar expansion of the charge distribution of the interacting subsystems which usually is cut off in the first term

12


which is different from zero. If both the solute and the solvent are considered to be formed by neutral polar molecules (with a permanent dipolar moment different from zero), due to an asymmetric distribution of its charges, the electric interaction of the type dipole-dipole will normally be the most important term in the electrostatic interaction. The intensity of this interaction will depend on the relative orientation of the dipoles. If the molecular rotation is not restricted, we must consider the weighted average over different orientations E d −d = −

2 µ 12µ 22 3 (4πε) 2 kTr 6

[2.1.3]

where: µ i, µ j k ε T r

dipole moments Boltzmann constant dielectric constant absolute temperature intermolecular distance

The most stable orientation is the antiparallel, except in the case that the molecules in play are very voluminous. Two dipoles in rapid thermal movement will be orientated sometimes in a way such that they are attracted and at other times in a way that they are repelled. On the average, the net energy turns out to be attractive. It also has to be borne in mind that the thermal energy of the molecules is a serious obstacle for the dipoles to be oriented in an optimum manner. The average potential energy of the diFigure 2.1.1. The dipoles of two molecules can approach one another under an infinite variety of attractive orienta- pole-dipole interaction, or of orientation, is, tions, among which these two extreme orientations stand therefore, very dependent on the temperaout. ture. In the event that one of the species involved were not neutral (for example an anionic or cationic solute) the predominant term in the series which gives the electrostatic interaction will be the ion-dipole which is given by the expression: E i −d = −

q i2 µ 2j 6( 4πε) 2 kTr 4

[2.1.4]

2.1.2.2 Polarization If we dissolve a polar substance in a nonpolar solvent, the molecular dipoles of the solute are capable of distorting the electronic clouds of the solvent molecules inducing the appearance in these of new dipoles. The dipoles of solute and those induced will line up and will be attracted and the energy of this interaction (also called interaction of polarization or induction) is:


E d − id = −

α j µ i2

13

[2.1.5]

( 4πε) 2 r 6

where: µi αj r

dipole moment polarizability intermolecular distance

In a similar way, the dissolution of an ionic substance in a nonpolar solvent also will occur with the induction of the dipoles in the molecules of the solvent by the solute ions. These equations make reference to the interactions between two molecules. Because the polarization energy (of the solute or of the solvent) is not pairwise additive magnitude, the consideration of a third molecule should be carried out simultaneously, it being impossible to decompose the interaction of the three bodies in a sum of the interactions of two bodies. The interactions between molecules in solution are different from those which take place between isolated molecules. For this reason, the dipolar moment of a molecule may vary considerably from the gas phase to the solution, and will depend in a complicated fashion on the interactions which may take place between the molecule of solute and its specific surroundings of molecules of solvent. 2.1.2.3 Dispersion Even when solvent and solute are constituted by nonpolar molecules, there is interaction between them. It was F. London who was first to face up to this problem, for which reason these forces are known as London’s forces, but also as dispersion forces, charge-fluctuations forces or electrodynamic forces. Their origin is as follows: when we say that a substance is nonpolar we are indicating that the distribution of the charges of its molecules is symmetrical throughout a wide average time span. But, without doubt, in an interval of time sufficiently restricted the molecular movements generate displacements of their charges which break that symmetry giving birth to instantaneous dipoles. Since the orientation of the dipolar moment vector is varying constantly due to the molecular movement, the average dipolar moment is zero, which does not prevent the existence of these interactions between momentary dipoles. Starting with two instantaneous dipoles, these will be oriented to reach a disposition which will favor them energetically. The energy of this dispersion interaction can be given, to a first approximation, by: E disp = −

3I i I j

αi α j

2( 4πε) (I i + I j ) r 6 2

[2.1.6]

where: Ii, Ij αi, αj r

ionization potentials polarizabilities intermolecular distance

From equation [2.1.6] it becomes evident that dispersion is an interaction which is more noticeable the greater the volume of molecules involved. The dispersion forces are often more intense than the electrostatic forces and, in any case, are universal for all the atoms and molecules, given that they are not seen to be subjected to the requirement that permanent dipoles should exist beforehand. These forces are responsible for the aggregation of the substances which possess neither free charges nor permanent dipoles, and are also the

14


protagonists of phenomena such as surface tension, adhesion, flocculation, physical adsorption, etc. Although the origin of the dispersion forces may be understood intuitively, this is of a quantum mechanical nature. 2.1.2.4 Repulsion Between two molecules where attractive forces are acting, which could cause them to be superimposed, it is evident that also repulsive forces exist which determine the distance to which the molecules (or the atoms) approach one another. These repulsive forces are a consequence of the overlapping of the electronic molecular clouds when these are nearing one another. These are also known as steric repulsion, hard core repulsion or exchange repulsion. They are forces of short range which grow rapidly when the molecules which interact approach one another, and which enter within the ambit of quantum mechanics. Throughout the years, different empirical potentials have been obtained with which the effect of these forces can be reproduced. In the model hard sphere potential, the molecules are supposed to be rigid spheres, such that the repulsive force becomes suddenly infinite, after a certain distance during the approach. Mathematically this potential is:  σ E rep =   r 

∞

[2.1.7]

where: r σ

intermolecular distance hard sphere diameter

Other repulsion potentials are the power-law potential:  σ E rep =   r 

n

[2.1.8]

where: r n σ

intermolecular distance integer, usually between 9 and 16 sphere diameter

and the exponential potential:  r E rep = C exp  −  σ0

  

[2.1.9]

where: r C σo

intermolecular distance adjustable constant adjustable constant

These last two potentials allow a certain compressibility of the molecules, more in consonance with reality, and for this reason they are also known as soft repulsion. If we represent the repulsion energy by a term proportional to r-12, and given that the energy of attraction between molecules decreases in proportion to r-6 at distances above the molecular diameter, we can obtain the total potential of interaction: E = −Ar −6 + Br −12

[2.1.10]


15

where: r A B

intermolecular distance constant constant

which receives the name of potential “6-12” or potential of Lennard-Jones,23 widely used for its mathematical simplicity (Figure 2.1.3) 2.1.2.5 Specific interactions Water, the most common liquid, the “universal solvent”, is just a little “extraordinary”, and this exceptional nature of the “liquid element” is essential for the world which has harbored us to keep on doing so. It is not normal that a substance in its solid state should be less dense than in the liquid, but if one ill-fated day a piece of ice spontaneously stopped floating on liquid water, all Figure 2.1.2. Hard-sphere repulsion (a) and soft repulwould be lost, the huge mass of ice which is sion (b) between two atoms. floating in the colder seas could sink thus raising the level of water in the oceans. For a liquid with such a small molecular mass, water has melting and boiling temperatures and a latent heat of vaporization which are unexpectedly high. Also unusual are its low compressibility, its high dipolar moment, its high dielectric constant and the fact that its density is maximum at 4 ºC. All this proves that water is an extraordinarily complex liquid in which the intermolecular forces exhibit specific interactions, the so-called hydrogen bonds, about which it is Figure 2.1.3. Lennard-Jones potential between two atnecessary to know more. oms. Hydrogen bonds appear in substances where there is hydrogen united covalently to very electronegative elements (e.g., F, Cl, O and N), which is the case with water. The hydrogen bond can be either intermolecular (e.g., H2O) or intramolecular (e.g., DNA). The protagonism of hydrogen is due to its small size and its tendency to become positively polarized, specifically to the elevated density of the charge which accumulates on the mentioned compounds. In this way, hydrogen is capable, such as in the case of water, of being doubly bonded: on the one hand it is united covalently to an atom of oxygen belonging to its molecule and, on the other, it electrostatically attracts another atom of oxygen belonging to another molecule, so strengthening the attractions between molecules. In this way, each atom of oxygen of a molecule of water can take part in four links with four more molecules of water, two of these links being through the hydrogen atoms covalently united to it and the other two links through hydrogen bonds thanks to the two pairs of solitary electrons which it

16


possesses. The presence of the hydrogen bonds together with this tetrahedric coordination of the molecule of water constitute the key to explaining its unusual properties. The energy of this bond (10-40 KJ/mol) is found to be between that corresponding to the van der Waals forces (~1 KJ/mol) and that corresponding to the simple covalent bond (200-400 KJ/mol). An energetic analysis of the hydrogen bond interaction shows that the leading term is the electrostatic one which explains that strong hydrogen bonds are found between hydrogen atoms with a partial positive charge and Figure 2.1.4. Tetrahedric structure of water in a crystal of a basic site. The second term in the energy ice. The dotted lines indicate the hydrogen bridges. decomposition of the hydrogen bond interaction is the charge transfer.24 The hydrogen bonds are crucial in explaining the form of the large biological molecules, such as the proteins and the nucleic acids, as well as how to begin to understand more particular chemical phenomena.25 Those solutes which are capable of forming hydrogen bonds have a well known affinity for the solvents with a similar characteristic, which is the case of water. The formation of hydrogen bonds between solute molecules and those of the solvent explains, for example, the good solubility in water of ammonia and of the short chain organic acids. 2.1.2.6 Hydrophobic interactions On the other hand, those nonpolar solutes which are not capable of forming hydrogen bonds with water, such as the case of the hydrocarbons, interact with it in a particular way. Let us imagine a molecule of solute incapable of forming hydrogen bonds in the midst of the water. Those molecules of water which come close to that molecule of solute will lose some or all of the hydrogen bonds which they were sharing with the other molecules of water. This obliges the molecules of water which surround those of solute to arrange themselves in space so that there is a loss of the least number of hydrogen bonds with other molecules of water. Evidently, this rearrangement (solvation or hydration) of the water molecules around the nonpolar molecule of solute will be greatly conditioned by the form and the size of this latter. All this amounts to a low solubility of nonpolar substances in water, which is known as the hydrophobic effect. If we now imagine not one but two nonpolar molecules in the midst of the water, it emerges that the interaction between these two molecules is greater when they are interacting in a free space. This phenomenon, also related to the rearrangement of the molecules of water around those of the solute, receives the name of hydrophobic interaction. The hydrophobic interaction term is used to describe the tendency of non-polar groups or molecules to aggregate in water solution.26 Hydrophobic interactions are believed to play a very important role in a variety of processes, specially in the behavior of proteins in aqueous media. The origin of this solvent-induced interactions is still unclear. In 1945 Frank and Evans27 proposed the so-called iceberg model where emphasis is made on the enhanced local structure of water around the non-polar solute. However, computational studies and ex-


17

perimental advances have yielded increasing evidence against the traditional interpretation,28 and other alternative explanations, such as the reduced freedom of water molecules in the solvation shell,29 have emerged. To understand the hydrophobic interaction at the microscopic level molecular simulations of non-polar compounds in water have been carried out.30 The potential of mean force between two non-polar molecules shows a contact minimum with an energy barrier. Computer simulations also usually predict the existence of a second solvent-separated minimum. Although molecular simulations provide valuable microscopic information on hydrophobic interactions they are computationally very expensive, specially for large systems, and normally make use of oversimplified potentials. The hydrophobic interaction can also be alternatively studied by means of continuum models.31 Using this approach a different but complementary view of the problem has been obtained. In the partition energy scheme used in the continuum models (see below and Chapter 8) the cavitation free energy (due to the change in the solvent-solvent interactions) is the most important contribution to the potential of mean force between two non-polar solutes in aqueous solution, being responsible for the energy barrier that separates the contact minimum. The electrostatic contribution to the potential of mean force for two non-polar molecules in water is close to zero and the dispersion-repulsion term remains approximately constant. The cavitation free energy only depends on the surface of the cavity where the solute is embedded and on the solvent physical properties (such as the surface tension and density). 2.1.3 MODELLING OF SOLVENT EFFECTS A useful way of understanding the interaction between the molecules of solvent and those of solute can be done by reproducing it by means of an adequate model. This task of imitation of the dissolution process usually goes beyond the use of simple and intuitive structural models, such as “stick” models, which prove to be very useful both in labors of teaching as in those of research, and frequently require the performance of a very high number of complex mathematical operations. Even though, in the first instance, we could think that a solution could be considered as a group of molecules united by relatively weak interactions, the reality is more complex, especially if we analyze the chemical reactivity in the midst of a solvent. The prediction of reaction mechanisms, the calculation of reaction rates, the obtaining of the structures of minimum energy and other precise aspects of the chemical processes in solution require the support of models with a very elaborated formalism and also of powerful computers. Traditionally, the models which permit the reproduction of the solute-solvent interactions are classified into three groups:32 i Those based on the simulation of liquids by means of computers. ii Those of continuum. iii Those of the supermolecule type. In the models classifiable into the first group, the system analyzed is represented by means of a group of interacting particles and the statistical distribution of any property is calculated as the the average over the different configurations generated in the simulation. Especially notable among these models are those of Molecular Dynamics and those of the Monte Carlo type. The continuum models center their attention on a microscopic description of the solute molecules, whilst the solvent is globally represented by means of its macroscopic properties, such as its density, its refractive index, or its dielectric constant.

18


Finally, the supermolecule type models restrict the analysis to the interaction among just a few molecules described at a quantum level which leads to a rigorous treatment of their interactions but does not allow us to have exact information about the global effect of the solvent on the solute molecules, which usually is a very long range effect. The majority of these models have their origin in a physical analysis of the solutions but, with the passage of time, they have acquired a more chemical connotation, they have centered the analysis more on the molecular aspect. As well as this, recourse is more and more being made to combined strategies which use the best of each of the methods referred to in pursuit of a truer reproduction of the solute-disolvente interactions. Specially useful has been shown to be a combination of the supermolecular method, used to reproduce the specific interactions between the solute and one or two molecules of the solvent, with those of continuum or of simulation, used to reproduce the global properties of the medium. 2.1.3.1 Computer simulations Obtaining the configuration or the conformation of minimum energy of a system provides us with a static view of this which may be sufficient to obtain many of its properties. However, direct comparison with experiments can be strictly be done only if average thermodynamic properties are obtained. Simulation methods are designed to calculate average properties of a system over many different configurations which are generated for being representative of the system behavior. These methods are based in the calculation of average properties as a sum over discrete events: N

F = ∫ dR1KdRn P(R )F (R ) ≈ ∑ Pi Fi

[2.1.11]

i =1

Two important difficulties arise in the computation of an average property as a sum. First, the number of molecules that can be handled in a computer is of the order of a few hundred. Secondly, the number of configurations needed to reach the convergency in the sum can be too great. The first problem can be solved by different computational strategies, such as the imposition of periodic boundary conditions.33 The solution of the second problem differs among the main used techniques in computer simulations. The two techniques most used in the dynamic study of the molecular systems are the Molecular Dynamics, whose origin dates back34 to 1957, and the Monte Carlo methods, which came into being following the first simulation of fluids by computer,35 which occurred in 1952. Molecular Dynamics In the Molecular Dynamic simulations, generation of new system configurations or sequence of events is made following the trajectory of the system which is dictated by the equations of motion. Thus, this methods leads to the computation of time averages and permits the calculation of not only equilibrium but also transport properties. Given a configuration of the system, a new configuration is obtained moving the molecules according to the total force exerted on them: md 2 R j dt

2

n

= −∑ ∇ j E (R jk ) k =1

[2.1.12]


19

If we are capable of integrating the equations of movement of all the particles which constitute a system, we can find their paths and velocities and we can evaluate the properties of the system in determined time intervals. Thus, we can find how the system being studied evolved as time moves forward. In the first simulations by Molecular Dynamics of a condensed phase,34 use was made of potentials as simple as the hard sphere potential, under which the constituent particles move in a straight line until colliding elastically. The collisions happen when the separation between the centers of the spheres is equal to the diameter of the sphere. After each collision, the new velocities are obtained by making use of the principle of conservation of the linear moments. But a chemical system requires more elaborate potentials under which the force, which at every instant acts between two atoms or molecules, changes in relation to the variation of the distance between them. This obliges us to integrate the equations of movement of the system in very small time intervals, in which it is assumed that the force which acts on each atom or molecule is constant, generally lying between 1 and 10 femtoseconds. For each of these intervals, the positions and velocities of each of the atoms is calculated, after which they are placed in their new positions and, once again, the forces are evaluated to obtain the parameters of a new interval, and so on, successively. This evolution in time, which usually requires the evaluation of hundreds of thousands of intervals of approximately 1 femtosecond each, allows us to know the properties of the system submitted to study during the elapse of time. In fact, the task commences by fixing the atoms which make up the system being studied in starting positions, and later move them continuously whilst the molecules being analysed rotate, the bond angles bend, the bonds vibrate, etc., and during which the dispositions of the atoms which make up the system are tabulated at regular intervals of time, and the energies and other properties which depend on each of the conformations, through which the molecular system makes its way with the passage of time, are evaluated. Molecular Dynamics is Chemistry scrutinized each femtosecond. Monte Carlo methods The first simulation by computer of a molecular system was carried out using this method. It consists of generating configurations of a system introducing random changes in the position of its constituents. In order to obtain a good convergence in the sequence of configurations, Metropolis et al.35 suggested an interesting approach. This approach avoids the generation of a very long random configurations as follows: instead of choosing random configurations and then weighing them according to the Boltzmann factor, one generates configurations with a probability equal to the Boltzmann factor and afterwards weigh them evenly. For this purpose once a new configuration is generated the difference in the potential energy with respect to the previous one is computed (∆U) and a random number 0 ≤ r ≤ 1 is selected. If the Boltzmann factor exp(-∆U/kT)>r then the new configuration is accepted, if exp(-∆U/kT) R CH OH > R C OH R''

The protonation free energies of MeOH to t-ButOH have been calculated in gas phase and with a continuum model of the solvent.61 It has been shown that in this case continuum models gives solvation energies which are good enough to correctly predict the acidity ordering of alcohols in solution. Simple electrostatic arguments based on the charge delocalization concept, were used to rationalize the progressive acidity of the alcohols when hydrogen atoms are substituted by methyl groups in the gas phase, with the effect on the solution energies being just the opposite. Thus, both the methyl stabilizing effect and the electrostatic interaction with the solvent can explain the acid scale in solution. As both terms are related to the molecular size, this explanation could be generalized for acid and base equilibria of homologous series of organic compounds: AH ⇔ A- + H+ B + H+ ⇔ BH+ In vacuo, as the size becomes greater by adding methyl groups, displacement of the equilibria takes place toward the charged species. In solution, the electrostatic stabilization is lower when the size increases, favoring the displacement of the equilibria toward the neutral species. The balance between these two tendencies gives the final acidity or basicity ordering in solution. Irregular ordering in homologous series are thus not unexpected taking into account the delicate balance between these factors in some cases.62 2.1.4.2 Solvent effects on the rate of chemical reactions When a chemical reaction takes place in the midst of a solution this is because, prior to this, the molecules of the reactants have diffused throughout the medium until they have met. This prior step of the diffusion of the reactants can reach the point of conditioning the performance of the reaction, especially in particularly dense and/or viscous surroundings. This is the consequence of the liquid phase having a certain microscopic order which, although


29

much less than that of the solid state, is not depreciable. Thus, in a solution, each molecule of solute finds itself surrounded by a certain number of molecules of solvent which envelope it forming what has been denominated as the solvent cage. Before being able to escape from the solvent cage each molecule of solute collides many times with the molecules of solvent which surround it. In the case of a dilute solution of two reactants, A and B, their molecules remain for a certain time in a solvent cage. If the time needed to escape the solvent cage by the molecules A and B is larger than the time needed to suffer a bimolecular reaction, we can say that this will not find itself limited by the requirement to overcome an energetic barrier, but that the reaction is controlled by the diffusion of the reactants. The corresponding reaction rate will, therefore, have a maximum value, known as diffusion-controlled rate. It can be demonstrated that the diffusion-limited bimolecular rate constants are of the order of 1010-1011 M-1s-1, when A and B are ions with opposite charges.63 For this reason, if a rate constant is of this order of magnitude, we must wait for the reaction to be controlled by the diffusion of the reactants. But, if the rate constant of a reaction is clearly less than the diffusion-limited value, the corresponding reaction rate is said to be chemically controlled. Focusing on the chemical aspects of the reactivity, the rupture of bonds which goes along with a chemical reaction usually occurs in a homolytic manner in the gas phase. For this reason, the reactions which tend to prevail in this phase are those which do not involve a separation of electric charge, such as those which take place with the production of radicals. In solution, the rupture of bonds tends to take place in a heterolytic manner, and the solvent is one of the factors which determines the velocity with which the process takes place. This explains that the reactions which involve a separation or a dispersion of the electric charge can take place in the condensed phase. The effects of the solvent on the reactions which involve a separation of charge will be very drawn to the polar nature of the state of transition of the reaction, whether this be a state of dipolar transition, isopolar or of the free-radical type. The influence of the solvent, based on the electric nature of the substances which are reacting, will also be essential, and reactions may occur between neutral nonpolar molecules, between neutral dipolar molecules, between ions and neutral nonpolar molecules, between ions and neutral polar molecules, ions with ions, etc. Moreover, we should bear in mind that alongside the non specific solute-solvent interactions (electrostatic, polarization, dispersion and repulsion), specific interactions may be present, such as the hydrogen bonds. Table 2.1.3. Relative rate constants of the Menschutkin reaction between triethylamine and iodoethane in twelve solvents at 50oC. In 1,1,1-trichloroethane the rate constant is 1.80×10-5 l mol-1 s-1. Data taken from reference 40 Solvent

Relative rate constant

1,1,1-Trichloroethane

1

Solvent

Relative rate constant

Acetone

17.61

Chlorocyclohexane

1.72

Cyclohexanone

18.72

Chlorobenzene

5.17

Propionitrile

33.11

Chloroform

8.56

Benzonitrile

42.50

1,2-Dichlorobenzene

10.06

30


Figure 2.1.12. Models of the reaction studied for the addition of azide anion to tetrafuranosides. The experimental product ratio is also given.

The influence of the solvent on the rate at which a chemical reaction takes place was already made clear, in the final stages of the XIX century, with the reaction of Menschutkin between tertiary amines and primary haloalkanes to obtain quaternary ammonium salts.64 The reaction of Menschutkin between triethylamine and iodoethane carried out in different media shows this effect (Table 2.1.3): 2.1.4.3 Example of application: addition of azide anion to tetrafuranosides The capacity of the solvent to modify both the thermodynamic and also the kinetic aspects of a chemical reaction are observed in a transparent manner on studying the stationary structures of the addition of azide anion to tetrafuranosides, particularly: methyl 2,3-dideoxy-2,3-epimino-α-L-erythrofuranoside (I), methyl 2,3-anhydro-α-L-erythrofuranoside (II), and 2,3-anhydro-β-L-erythrofuranoside (III). An analysis with molecular orbital methods at the HF/3-21G level permits the potential energy surface in vacuo to be characterized, to locate the stationary points and the possible reaction pathways.65 The effect of the solvent can be implemented with the aid of a polarizable continuum model. Figure 2.1.12 shows the three tetrafuranosides and the respective products obtained when azide anion attacks in C3 (P1) or in C4 (P2). The first aim of a theoretical study of a chemical reaction is to determine the reaction mechanism that corresponds to the minimum energy path that connects the minima of reactant and products and passes through the transition state (TS) structures on the potential en-


31

Figure 2.1.13. Representation of the stationary points (reactants, reactant complex, transition states, and products) for the molecular mechanism of compound I. For the TS´s the components of the transition vectors are depicted by arrows.

ergy surface. In this path the height of the barrier that exists between the reactant and TS is correlated to the rate of each different pathway (kinetic control), while the relative energy of reactants and products is correlated to equilibrium parameters (thermodynamic control). The second aim is how the solute-solvent interactions affect the different barrier heights and relative energies of products, mainly when charged or highly polar structures appear along the reaction path. In fact, the differential stabilization of the different stationary points in the reaction paths can treat one of them favorably, sometimes altering the relative energy order found in vacuo and, consequently, possibly changing the ratio of products of the reaction. An analysis of the potential energy surface for the molecular model I led to the location of the stationary points showed in Figure 2.1.13. The results obtained for the addition of azide anion to tetrafuranosides with different molecular models and in different solvents can be summarized as follows:65 • For compound I, in vacuo, P1 corresponds to the path with the minimum activation energy, while P2 is the more stable product (Figures 2.1.14 and 2.1.15). When the solvent effect is included, P1 corresponds to the path with the minimum activation energy and it is also the more stable product. For compound II, in vacuo, P2 is the more stable product and also presents the smallest activation energy. The inclusion of the solvent effect in this case changes the order of products and transition states stability. For compound III, P1 is the more stable product and presents the smallest activation energy both in vacuo and in solution. • A common solvent effect for the three reactions is obtained: as far as the dielectric constant of the solvent is augmented, the energy difference between P1 and P2

32


Figure 2.1.14. Schematic representation of the relative energies (in Kcal/mol) for the products P1 and P2, in vacuo (ε = 1), DMF ( ε = 36.7), or EtOH (ε = 24.3), and in water (ε = 78.4).

Figure 2.1.15. Schematic representation of the relative energies of activation (in Kcal/mol) for the transition states TS1 and TS2, in vacuo (ε = 1), DMF (ε = 36.7), or EtOH (ε = 24.3), and in water (ε = 78.4).

increases, favoring thermodynamically the path leading to P1. On the other hand, an opposite influence is evident in the case of the transition states, so an increase of the dielectric constant kinetically favors the path leading to P2. All this data makes evident the crucial role which the solvent plays both in the thermodynamics and in the kinetics of the chemical reaction analysed. 2.1.5 SOLVENT CATALYTIC EFFECTS Beyond the solvent as merely making possible an alternative scenery to the gas phase, beyond its capacity to alter the thermodynamics of a process, the solvent can also act as a catalyst of some reactions, and can reach the point of altering the mechanism by which the reaction comes about. An example of a reaction in which the solvent is capable of altering the mechanism through which the reaction takes place is that of Meyer-Schuster, which is much utilized in organic synthesis.66-71 This conFigure 2.1.16. Reaction of Meyer-Schuster. sists of the isomerization in an acid medium


33

of secondary and tertiary α-acetylenic alcohols to carbonylic α , β-unsaturated compounds (Figure 2.1.16). Its mechanism consists of three steps (Figure 2.1.17). The first one is the protonation of the oxygen atom. The second, which determines the reaction rate, is that in which the 1.3 shift from the protonated hydroxyl group is produced through the triple bond to give way to the structure of alenol. The last stage corresponds to the deprotonation of the alenol, producing a keto-enolic tautomerism which displaces towards the ketonic form. For the step which limits the Figure 2.1.17. Steps of the reaction of Meyer-Schuster. reaction rate (rate limiting step), three mechanisms have been proposed, two of which are intramolecular - denominated intramolecular, as such, and solvolytic - and the other intermolecular (Figure 2.1.18). The first of these implies a covalent bond between water and the atoms of carbon during the whole of the transposition. In the solvolytic mechanism there is an initial rupture from the O-C1 bond, followed by a nucleophilic attack of the H2O on the C3. Whilst the intermolecular mechanism corresponds to a nucleophilic attack of H2O on the terminal carbon C3 and the loss of the hydroxyl group protonated of the C1. The analysis of the first two mechanisms showed72 the solvolytic mechanism as the most favorable localizing itself during the reaction path to an alquinylic carbocation interacting electrostatically with a molecule of water. This fact has been supported by the experimental detection of alquinylic carbocations in solvolytic conditions. Things being like that, two alternatives remain for the slow stage of the Meyer-Schuster reaction, the solvolytic and the intermolecular mechanism, and it seems that the solvent has a lot to say in this. Although both mechanisms evolve in two steps, these are notably different. In the intermolecular mechanism, the first transition state can be described as an almost pure electrostatic interaction of the entrant molecule of water with the C3, whilst the C1 remains united covalently to the protonated hydroxyl group. This first transition state leads to a intermediate in which the two molecules of water are covalently bonded to the C1 and C3 atoms. The step from the intermediate to the product takes place through a second transition state, in which the C3 is covalently bonded to the molecule of entrant water and there is an electrostatic interaction of the other water molecule. In the mechanism which we call solvolytic, the first transition state corresponds to the pass from covalent to electrostatic interaction of the H2O united to the C1, that is to say, to a process of solvolysis, so, the water molecule remains interacting electrostatically with the carbons C1 and C3.

34


Figure 2.1.18. Three different mechanism for the rate-limiting step of the reaction of Meyer-Schuster.

On comparing the solvolytic and intermolecular processes a smaller potential barrier is observed for the latter, thus the solvent plays an active part in the Meyer-Schuster reaction, being capable of changing radically the mechanism through which this takes place. It seems clear that in the presence of aqueous solvents the nucleophilic attack on the C3 precedes the loss of the water (solvolysis): a lesser activation energy corresponds to the intermolecular process than to the solvolytic. Moreover, if we analyze the intermolecular mechanism we can verify that the solvent stabilizes both the reactants as well as the products by the formation of hydrogen bridges. Epilogue As a fish in the midst of the ocean, the reactants are usually found in the midst of a solution in our laboratory tests. In the same way as in the ocean where there is both danger and a heaven for the fish, in the internal scenery of a solution the chemical reactions can be speeded up or slowed down, favored thermodynamically or prejudiced. In this way, on passing from vacuum to a solution, the molecules of the reactants can experience alterations in their geometry, the distribution of their charges, or their energy, which can have an effect on the outcome of the reaction. In the preceding pages we have attempted to make clear these solute-solvent influences, and to achieve this we have plunged, hand in hand with theoretical chemistry, into the microscopic and recondite environment of the solutions. REFERENCES 1 2 3 4 5 6 7 8 9

Ch. Reichardt, Solvents and Solvent Effects in Organic Chemistry, VCH, Weinheim, 1988, p. 1. H. Metzger, Newton, Stahl, Boerhaave et la doctrine chimique, Alcan, Paris, 1930, pp. 280-289. M.M. Pattison, A History of Chemical Theories and Laws, John Wiley & Sons, New York, 1907, p. 381. H.M. Leicester, Panorama histórico de la química, Alhambra, Madrid, 1967, pp. 148, 149. (Transcription from: The Historical Background of Chemistry, John Wiley & Sons, New York.) J.L. Proust, J. Phys., 63, 369, 1806. M. Berthelot and L. Péan de Saint-Gilles, Ann. Chim. et Phys., Ser. 3, 65, 385 (1862); 66, 5 (1862); 68, 255 (1863). N. Menschutkin, Z. Phys. Chem., 6, 41 (1890). E. D. Hughes, C. K. Ingold, J. Chem. Soc., 244, 252 (1935). L. Claisen, Liebigs Ann. Chem., 291, 25 (1896).


10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

31 32 33 34 35 36 37 38 39 40 41 42

43

44

45 46 47

35

W. Wislicenus. Liebigs Ann. Chem., 291, 147 (1896). J. E. Bartmess, J. A. Scott, R. T. McIver Jr., J. Am. Chem. Soc., 101, 6056 (1979). L. M. Jackman, D. P. Kelly, J. Chem. Soc., (B) 102 (1970). H. C. Brown, H. Bartholomay, M. D. Taylor, J. Am. Chem. Soc., 66, 435 (1944). J. P. Briggs, R. Yamdagni, P. Kebarle, J. Am. Chem. Soc., 94, 5128 (1972). T. A. Leheman, M. M. Bursey, Ion Cyclotron Resonance Spectrometry, Wiley, New York, 1976. S. A. McLukey, D. Cameron, R. G. Cooks, J. Am. Chem. Soc., 103, 1313 (1981). B. K. Bohme, P. Fennelly, R. S: Hemsworth, H. J. Schiff, J. Am. Chem. Soc. b, 7512 (1973). J. E. Bartmess, R. T. McIver, Gas phase Ion Chemistry, M. T. Bowe ed., Ed., Academic Press, New York, 1979. J. W. Larson, T. B. McMahon, J. Am. Chem. Soc., 104, 6255 (1982). M. Meotner, L. W: Sieck, J. Am. Chem. Soc., 105, 2956 (1983). C. L. Brooks III, M. Karplus, Methods Enzymol., 127, 369 (1986). D. Beece, L. Eisenstein, H. Frauenfelder, D. Good, M. C. Marden, L. Reinisch, A. H. Reynolds, L. B. Sorensen, K. T. Yue, Biochemistry, 19, 5147 (1980). J.E. Lennard-Jones, Proc. Phys. Soc. London, 43, 461 (1931). J. P. Daudey, Int. J. Quantum Chem., b, 29 (1974); J. L. Rivail, Éléments de Chimie Quantique à l’usage des Chimistes, InterÉditions-CNRS Éditions, Paris (1994). C. Sandorfy, R. Buchet, L.S. Lussier, P. Ménassa and L. Wilson, Pure Appl. Chem., 58, 1115 (1986). J. N., Israelachvili, Intermolecular and surface forces, Academic Press, San Diego, 1989. W. Kauzmann, Adv. Protein Chem., 14, 1 (1959). H. S. Frank and M. W. Evans, J. Chem. Phys., 13, 507 (1945). W. Blokzilj and J. B. F. N. Engberts, Angew. Chem. Int. Ed. Engl., 32, 1545 (1993). B. Lee, Biopolymers, 24, 813 (1985), B. Lee, Biopolymers, 31, 993 (1991). W. L. Jorgensen, J. K. Bukner, S. Boudon and J. Tirado-Rives, Chem. Phys., 89, 3742 (1988); C. Pangali, M. Rao and B. J. Berne, J. Chem. Phys., 71, 2975(1979); D. van Belle and S. J. Wodak, J. Am. Chem. Soc., 115, 647 (1993); M. H. New and B. J. Berne, J. Am. Chem. Soc., 117, 7172 (1995); D. E. Smith, L. Zhang and A. D. J. Haymet, J. Am. Chem. Soc., 114, 5875 (1992); ); D. E. Smith and A. D. J. Haymet, J. Chem. Phys., 98, 6445(1994); L. X. Dang, J. Chem. Phys., 100, 9032(1994); T. Head-Gordon, Chem. Phys. Lett., 227, 215 (1994). J. Pitarch, V. Moliner, J. L. Pascual-Ahuir, E. Silla and I. Tuñçon , J. Phys. Chem., 100, 9955(1996). J. Tomasi and M. Persico, Chem. Rev., 94, 2027-2094 (1994). A. Rahman and F.M. Stillinger, J. Chem. Phys., 57, 3336 (1971). A. R. Leach, Molecular Modelling, Addison Wesley Longman, Singapore, 1997. B.J. Alder and T.E. Wainwright, J. Chem. Phys., 27, 1208-1209 (1957). N. Metropolis, A.W. Rosenbluth, M.N. Rosenbluth, A. Teller and E. Teller, J. Chem. Phys., 21, 1087 (1953) A. Warshel, M. Levitt, J. Mol. Biol., 103, 227-249 (1973). P. A. Bash, M. J. Field, M. Karplus, J. Am. Chem. Soc., 109, 8092-8094 (1987). M. Born, Z. Physik, 1, 45 (1920); J. G. Kirkwood, J. Chem. Phys., 2, 351 (1934); J. G. Kirkwood, F. H. Westheimer, J. Chem. Phys., 6, 506 (1936); L. Onsager, J. Am. Chem. Soc., 58, 1486 (1936). E. Scrocco, J. Tomasi, Top. Curr. Chem., 42, 97 (1973). J. L. Pascual-Ahuir, E. Silla, J. Tomasi, R. Bonaccorsi, J. Comput. Chem., 8, 778 (1987). D. Rinaldi, J. L. Rivail, N. Rguini, J. Comput. Chem., b, 676 (1992). I. Tuñón, E. Silla, J. Bertrán, J. Chem. Soc. Fraday Trans., 90, 1757 (1994). X. Assfeld, D. Rinaldi, AIP Conference Proceedings ECCC1, F. Bernardi, J. L. Rivail, Eds., AIP, Woodbury, New York, 1995. F. M. Floris, J. Tomasi, J. L. Pascual Ahuir, J. Comput. Chem., 12, 784 (1991); C. J. Cramer and D. G. Trhular, Science, 256, 213 (1992); V. Dillet, D. Rinaldi, J. G. Angyan and J. L. Rivail, Chem. Phys. Lett., 202, 18 (1993). I. Tuñón, E. Silla, J. L. Pascual-Ahuir, Prot. Eng., 5, 715 (1992); I. Tuñón, E. Silla, J. L. Pascual-Ahuir, Chem. Phys. Lett., 203, 289 (1993); R. B. Hermann, J. Phys. Chem., 76, 2754 (1972); K. A. Sharp, A. Nicholls, R. F: Fine, B. Honig, Science, 252, 106 (1991). L. H. Pearl, A. Honegger, J. Mol. Graphics, 1, 9 (1983); R. Voorintholt, M. T. Kosters, G. Vegter, G. Vriend, W. G. H. Hol, J. Mol. Graphics, 7, 243 (1989); J. B. Moon, W. J. Howe, J. Mol. Graphics, 7, 109 (1989); E. Silla, F. Villar, O. Nilsson, J.L. Pascual-Ahuir, O. Tapia, J. Mol. Graphics, 8, 168 (1990). B. Lee, F. M. Richards, J. Mol. Biol., 55, 379 (1971). F. M. Richards, Ann. Rev. Biophys. Bioeng., 6, 151 (1977). J. L. Pascual-Ahuir, E. Silla, I. Tuñón, J. Comput. Chem., 15, 1127-1138 (1994).

36

Koichiro Nakanishi

48

V. Moliner, J. Andrés, A. Arnau, E. Silla and I. Tuñón, Chem. Phys., 206, 57-61 (1996); R. Moreno, E. Silla, I. Tuñón and A. Arnau, Astrophysical J., 437, 532-539 (1994); A. Arnau, E. Silla and I. Tuñón, Astrophysical J. Suppl. Ser., 88, 595-608 (1993); A. Arnau, E. Silla and I. Tuñón, Astrophysical J., 415, L151-L154 (1993). M. D. Newton, S. J. Ehrenson, J. Am. Chem. Soc., 93, 4971 (1971). G. Alagona, R. Cimiraglia, U. Lamanna, Theor. Chim. Acta, 29, 93 (1973). J. E: del Bene, M. J. Frisch, J. A. Pople, J. Phys. Chem., 89, 3669 (1985). I. Tuñón, E. Silla, J. Bertrán, J. Phys. Chem., 97, 5547-5552 (1993). R. Car, M. Parrinello, Phys. Rev. Lett., 55, 2471 (1985). P. G. Jonsson and A. Kvick, Acta Crystallogr. B, 28, 1827 (1972). A. G. Csázár, Theochem., 346, 141-152 (1995). Y. Ding and K. Krogh-Jespersen, Chem. Phys. Lett., 199, 261-266 (1992). J. H. Jensen and M.S.J. Gordon, J. Am. Chem. Soc., 117, 8159-8170 (1995). F. R. Tortonda., J.L. Pascual-Ahuir, E. Silla, and I. Tuñón, Chem. Phys. Lett., 260, 21-26 (1996). T.N. Truong and E.V. Stefanovich, J. Chem. Phys., 103, 3710-3717 (1995). N. Okuyama-Yoshida et al., J. Phys. Chem. A, 102, 285-292 (1998). I. Tuñón, E. Silla, C. Millot, M. Martins-Costa and M.F. Ruiz-López, J. Phys. Chem. A, 102, 8673-8678 (1998). F. R. Tortonda, J.L. Pascual-Ahuir, E. Silla, I. Tuñón, and F.J. Ramírez, J. Chem. Phys., 109, 592-602 (1998). F.J. Ramírez, I. Tuñón, and E. Silla, J. Phys. Chem. B, 102, 6290-6298 (1998). I. Tuñón, E. Silla and J.L. Pascual-Ahuir, J. Am. Chem. Soc., 115, 2226 (1993). I. Tuñón, E. Silla and J. Tomasi, J. Phys. Chem., 96, 9043 (1992). K.A. Connors, Chemical Kinetics, VCH, New York, 1990, pp. 134-135. N. Menschutkin, Z. Phys. Chem., 6, 41 (1890); ibid., 34, 157 (1900). J. Andrés, S. Bohm, V. Moliner, E. Silla, and I. Tuñón, J. Phys. Chem., 98, 6955-6960 (1994). S. Swaminathan and K.V. Narayan, Chem. Rev., 71, 429 (1971). N.K. Chaudhuri and M. Gut, J. Am. Chem. Soc., 87, 3737 (1965). M. Apparu and R. Glenat, Bull. Soc. Chim. Fr., 1113-1116 (1968). S. A. Vartanyan and S.O. Babayan, Russ. Chem. Rev., 36, 670 (1967). L.I. Olsson, A. Claeson and C. Bogentoft, Acta Chem. Scand., 27, 1629 (1973). M. Edens, D. Boerner, C. R. Chase, D. Nass, and M.D. Schiavelli, J. Org. Chem., 42, 3403 (1977). J. Andrés, A. Arnau, E. Silla, J. Bertrán, and O. Tapia, J. Mol. Struct. Theochem., 105, 49 (1983); J. Andrés, E. Silla, and O. Tapia, J. Mol. Struct. Theochem., 105, 307 (1983); J. Andrés, E. Silla, and O. Tapia, Chem. Phys. Lett., 94, 193 (1983); J. Andrés, R. Cárdenas, E. Silla, O. Tapia, J. Am. Chem. Soc., 110, 666 (1988).

49

50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72

2.2 MOLECULAR DESIGN OF SOLVENTS Koichiro Nakanishi Kurashiki Univ. Sci. & the Arts, Okayama, Japan

2.2.1 MOLECULAR DESIGN AND MOLECULAR ENSEMBLE DESIGN Many of chemists seem to conjecture that the success in developing so-called high-functional materials is the key to recover social responsibility. These materials are often composed of complex molecules, contain many functional groups and their structure is of complex nature. Before establishing the final target compound, we are forced to consider many factors, and we are expected to minimize the process of screening these factors effectively. At present, such a screening is called “design”. If the object of screening is each molecule, then it is called “molecular design”. In similar contexts are available “material design”, “solvent design”, “chemical reaction design”, etc. We hope that the term “molecular ensemble design” could have the citizenship in chemistry. The reason for this is as below. Definition of “molecular design” may be expressed as to find out the molecule which has appropriate properties for a specific purpose and to predict accurately via theoretical ap-

2.2 Molecular design of solvents

37

proach the properties of the molecule. If the molecular system in question consists of an isolated free molecule, then it is “molecular design”. If the properties are of complex macroscopic nature, then it is “material design”. Problem remains in the intermediate between the above two. Because fundamental properties shown by the ensemble of molecules are not always covered properly by the above two types of design. This is because the molecular design is almost always based on quantum chemistry of free molecule and the material design relies too much on empirical factor at the present stage. When we proceed to molecular ensemble (mainly liquid phase), as the matter of fact, we must use statistical mechanics as the basis of theoretical approach. Unfortunately, statistical mechanics is not familiar even for the large majority of chemists and chemical engineers. Moreover, fundamental equations in statistical mechanics cannot often be solved rigorously for complex systems and the introduction of approximation becomes necessary to obtain useful results for real systems. In any theoretical approach for molecular ensemble, we must confront with so-called many-body problems and two-body approximations must be applied. Even in the frameworks of this approximation, our knowledge on the intermolecular interaction, which is necessary in statistical mechanical treatment is still poor. Under such a circumstance, numerical method should often be useful. In the case of statistical mechanics of fluids, we have Monte Carlo (MC) simulation based on the Metropolis scheme. All the static properties can be numerically calculated in principle by the MC method. Another numerical method to supplement the MC method should be the numerical integration of the equations of motion. This kind of calculation for simple molecular systems is called molecular dynamics (MD) method where Newton or Newton-Euler equation of motion is solved numerically and some dynamic properties of the molecule involved can be obtained. These two methods are invented, respectively, by the Metropolis group (MC, Metropolis et al., 1953)1 and Alder’s group (MD, Alder et al., 1957)2 and they are the molecular versions of computer experiments and therefore called now molecular simulation.3 Molecular simulation plays a central role in “molecular ensemble design”. They can reproduce thermodynamics properties, structure and dynamics of a group of molecules by using high speed supercomputer. Certainly any reasonable calculations on molecular ensemble need long computer times, but the advance in computer makes it possible that this problem becomes gradually less serious. Rather, the assignment is more serious with intermolecular interaction potential used. For simple molecules, empirical model potential such as those based on Lennard-Jones potential and even hard-sphere potential can be used. But, for complex molecules, potential function and related parameter value should be determined by some theoretical calculations. For example, contribution of hydrogen-bond interaction is highly large to the total interaction for such molecules as H2O, alcohols etc., one can produce semi-empirical potential based on quantum-chemical molecular orbital calculation. Molecular ensemble design is now complex unified method, which contains both quantum chemical and statistical mechanical calculations. 2.2.2 FROM PREDICTION TO DESIGN It is not new that the concept of “design” is brought into the field of chemistry. Moreover, essentially the same process as the above has been widely used earlier in chemical engineer-

38

Koichiro Nakanishi

ing. It is known as the prediction and correlation methods of physical properties, that is, the method to calculate empirically or semi-empirically the physical properties, which is to be used in chemical engineering process design. The objects in this calculation include thermodynamic functions, critical constant, phase equilibria (vapor pressure, etc.) for one-component systems as well as the transport properties and the equation of state. Also included are physical properties of two-components (solute + pure solvent) and even of three-components (solute + mixed solvent) systems. Standard reference, “The properties of Gases and Liquids; Their Estimation and Prediction”,4 is given by Sherwood and Reid. It was revised once about ten years interval by Reid and others. The latest 4th edition was published in 1987. This series of books contain excellent and useful compilation of “prediction” method. However, in order to establish the method for molecular ensemble design we need to follow three more stages. (1) Calculate physical properties of any given substance. This is just the establishment and improvement of presently available “prediction” method. (2) Calculate physical properties of model-substance. This is to calculate physical properties not for each real molecule but for “model”. This can be done by computer simulation. On this stage, compilation of model"substance data base will be important. (3) Predict real substance (or corresponding “model”) to obtain required physical properties. This is just the reverse of the stage (1) or (2). But, an answer in this stage is not limited to one particular substance. The scheme to execute these three stages for a large variety of physical properties and substances has been established only to a limited range. Especially, important is the establishment of the third stage, and after that, “molecular ensemble design” will be worth to discuss. 2.2.3 IMPROVEMENT IN PREDICTION METHOD Thus, the development of “molecular ensemble design” is almost completely future assignment. In this section, we discuss some attempts to improve prediction at the level of stage (1). It is taken for the convenience’s sake from our own effort. This is an example of repeated improvements of prediction method from empirical to molecular level. The diffusion coefficient D1 of solute 1 in solvent 2 at infinitely dilute solution is a fundamental property. This is different from the self-diffusion coefficient D0 in pure liquid. Both D1 and D0 are important properties. The classical approach to D1 can be done based on Stokes and Einstein relation to give the following equation D1η 2 = kT / 6πr1

[2.2.1]

where D1 at constant temperature T can be determined by the radius r1 of solute molecule and solvent viscosity η2 (k is the Boltzmann constant). However, this equation is valid only when the size of solvent molecule is infinitely small, namely, for the diffusion in continuous medium. It is then clear that this equation is inappropriate for molecular mixtures. The well known Wilke-Chang equation,5 which corrects comprehensively this point, can be used for practical purposes. However, average error of about 10% is inevitable in the comparison with experimental data.


39

One attempt6 to improve the agreement with experimental data is to use Hammond-Stokes relation in which the product D1η2 is plotted against the molar volume ratio of solvent to solute Vr. The slope is influenced by the following few factors, namely, (1) self-association of solvent, (2) asymmetricity of solvent shape, and (3) strong solute-solvent interactions. If these factors can be taken properly into account, the following equation is obtained. D1η 2 / T = K1 / V01/ 3 + K 2Vr / T

[2.2.2]

Here constants K1 and K2 contain the parameters coming from the above factors and V0 is the molar volume of solute. This type of equation [2.2.2] cannot be always the best in prediction, but physical image is clearer than with other purely empirical correlations. This is an example of the stage (1) procedure and in order to develop a stage (2) method, we need MD simulation data for appropriate model. 2.2.4 ROLE OF MOLECULAR SIMULATION We have already pointed out that statistical mechanical method is indispensable in “molecular ensemble design”. Full account of molecular simulation is given in some books3 and will not be reproduced here. Two types of approaches can be classified in applying this method. The first one makes every effort to establish and use as real as exact intermolecular interaction in MC and MD simulation. It may be limited to a specific type of compounds. The second is to use simple model, which is an example of so-called Occam’s razor. We may obtain a wide bird-view from there. In the first type of the method, intermolecular interaction potential is obtained based on quantum-chemical calculation. The method takes the following steps. (1) Geometry (interatomic distances and angles) of molecules involved is determined. For fundamental molecules, it is often available from electron diffraction studies. Otherwise, the energy gradient method in molecular orbital calculation can be utilized. (2) The electronic energy for monomer E1 and those for dimers of various mutual configurations E2 are calculated by the so-called ab initio molecular orbital method, and the intermolecular energy E2-2E1 is obtained. (3) By assuming appropriate molecular model and semi-empirical equation, parameters are optimized to reproduce intermolecular potential energy function. Representative example of preparation of such potential energy function, called now ab initio potential would be MCY potential for water-dimer by Clementi et al.7 Later, similar potentials have been proposed for hetero-dimer such as water-methanol.8 In the case of such hydrogen-bonded dimers, the intermolecular energy E2-2E1 can be fairly large value and determination of fitted parameters is successful. In the case of weak interaction, ab initio calculation needs long computer time and optimization of parameter becomes difficult. In spite of such a situation, some attempts are made for potential preparation, e.g., for benzene and carbon dioxide with limited success. To avoid repeated use of long time MO calculation, Jorgensen et al.9 has proposed MO-based transferable potential parameters called TIPS potential. This is a potential ver-

40

Koichiro Nakanishi

Figure 2.2.1. A scheme for the design of molecular ensembles.

sion of additivity rule, which has now an empirical character. It is however useful for practical purposes. 2.2.5 MODEL SYSTEM AND PARADIGM FOR DESIGN The method described above is so to speak an orthodox approach and the ability of present-day’s supercomputer is still a high wall in the application of molecular simulation. Then the role of the second method given in the last section is highly expected. It is the method with empirical potential model. As the model, the approximation that any molecule can behave as if obeying Lennard-Jones potential seems to be satisfactory. This (one-center) LJ model is valid only for rare gases and simple spherical molecules. But this model may also be valid for other simple molecules as a zeroth approximation. We may also use two-center LJ model where interatomic interactions are concentrated to the major two atoms in the molecule. We expect that these one-center and two-center LJ models will play a role of Occam’s razor. We propose a paradigm for physical properties prediction as shown in Figure 2.2.1. This corresponds to the stage (2) and may be used to prepare the process of stage (3), namely, the molecular ensemble design for solvents. Main procedures in this paradigm are as follows; we first adopt target molecule or mixture and determine their LJ parameters. At present stage, LJ parameters are available only for limited cases. Thus we must have method to predict effective LJ parameters.


41

For any kinds of mixtures, in addition to LJ parameters for each component, combining rule (or mixing rule) for unlike interaction should be prepared. Even for simple liquid mixtures, conventional Lorentz-Berthelot rule is not good answer. Once potential parameters have been determined, we can start calculation downward following arrow in the figure. The first key quantity is radial distribution function g(r) which can be calculated by the use of theoretical relation such as Percus-Yevick (PY) or Hypernetted chain (HNC) integral equation. However, these equations are an approximations. Exact values can be obtained by molecular simulation. If g(r) is obtained accurately as functions of temperature and pressure, then all the equilibrium properties of fluids and fluid mixtures can be calculated. Moreover, information on fluid structure is contained in g(r) itself. On the other hand, we have, for non-equilibrium dynamic property, the time correlation function TCF, which is dynamic counterpart to g(r). One can define various TCF’s for each purpose. However, at the present stage, no extensive theoretical relation has been derived between TCF and φ(r). Therefore, direct determination of self-diffusion coefficient, viscosity coefficient by the molecular simulation gives significant contribution in dynamics studies. Concluding Remarks Of presently available methods for the prediction of solvent physical properties, the solubility parameter theory by Hildebrand10 may still supply one of the most accurate and comprehensive results. However, the solubility parameter used there has no purely molecular character. Many other methods are more or less of empirical character. We expect that the 21th century could see more computational results on solvent properties. REFERENCES 1 2 3 4 5 6 7 8 9 10

N. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller and E. Teller, J. Chem. Phys., 21, 1087 (1953). B. J. Alder and T. E. Wainwright, J. Chem. Phys., 27, 1208 (1957). (a) M. P. Allen and D. J. Tildesley, Computer Simulation of Liquids, Clarendon Press, Oxford, 1987. (b) R. J. Sadus, Molecular Simulation of Fluids, Elsevier, Amsterdam, 1999. R. C. Reid, J. M. Prausnitz and J. E. Poling, The Properties of Gases and Liquids; Their Estimation and Prediction, 4th Ed., McGraw-Hill, New York, 1987. C. R. Wilke and P. Chang, AIChE J., 1, 264 (1955). K. Nakanishi, Ind. Eng. Chem. Fundam., 17, 253 (1978). O. Matsuoka, E. Clementi and M. Yoshimine, J. Chem. Phys., 60, 1351 (1976). S. Okazaki, K. Nakanishi and H. Touhara, J. Chem. Phys., 78, 454 (1983). W. L. Jorgensen, J. Am. Chem. Soc., 103, 345 (1981). J. H. Hildebrand and R. L. Scott, Solubility of Non-Electrolytes, 3rd Ed., Reinhold, New York, 1950.

APPENDIX PREDICTIVE EQUATION FOR THE DIFFUSION COEFFICIENT IN DILUTE SOLUTION Experimental evidence is given in Figure 2.2.2 for the prediction based on equation [2.2.1]. The diffusion coefficient D0 of solute A in solvent B at an infinite dilution can be calculated using the following equation:  9.97 × 10 −8 2.40 × 10 −8 A S V B B B + D0 =   [I V ]1/ 3 I S V A A A  A A

T   ηB 

[2.2.3]

42

Koichiro Nakanishi

Figure 2.2.2. Hammond-Stokes plot for diffusion of iodine and carbon tetrachloride in various solvents at 298.15K. o, ,: D1η2, n: (D1)selfη2, : D1η2 at the Stokes-Einstein limit. Perpendicular lines connect two or more data for the same solvent from different sources. [Adapted, by permission, from K. Nakanishi, Bull. Chem. Soc. Japan, 51, 713 (1978).

where D0 is in cm2 s-1. VA and VB are the liquid molar volumes in cm3 mol-1 of A and B at the temperature T in K, and the factors I, S, and A are given in original publications,4,6 and η is the solvent viscosity, in cP. Should the pure solute not be a liquid at 298 K, it is recommended that the liquid volume at the boiling point be obtained either from data or from correlations.4 Values of D0 were estimated for many (149) solute-solvent systems and average error was 9.1 %.

2.3 BASIC PHYSICAL AND CHEMICAL PROPERTIES OF SOLVENTS George Wypych ChemTec Laboratories, Inc., Toronto, Canada

This section contains information on the basic relationships characterizing the physical and chemical properties of solvents and some suggestions regarding their use in solvent evaluation and selection. The methods of testing which allow us to determine some of physical and chemical properties are found in Chapter 15. The differences between solvents of various chemical origin are discussed in Chapter 3, Section 3.3. The fundamental relationships in this chapter and the discussion of different groups of solvents are based on extensive CD-ROM database of solvents which can be obtained from ChemTec Publishing. The database has 110 fields which contain various data on solvent properties which are discussed below. The database can be searched by the chemical name, empirical formula, molecular weight, CAS number and property. In the first case, full information on a particular solvent

2.3 Basic physical and chemical properties

43

Hildebrand solubility parameter,δ, cal cm

-3

is returned. In the second case, a list of solvents and their values for the selected property are given in tabular form in ascending order of the property in question. 2.3.1 MOLECULAR WEIGHT AND MOLAR VOLUME The molecular weight of a solvent is a standard but underutilized component of the information on properties of solvents. Many solvent properties depend directly on their molecular weights. The hypothesis of Hildebrand-Scratchard states that solvent-solute interaction occurs when solvent and polymer segment have similar molecular weights. This is related to the hole theory according to which a solvent occupying a certain volume leaves the same volume free when it is displaced. This free volume should be sufficient to fit the polymer segment which takes over the position formerly occupied by the solvent molecule. Based on this same principle, the diffusion coefficient of a solvent depends on its molecular mass (see equations [6.2] and [6.3]). As the molecular weight of a solvent increases its diffusion rate also increases. If there were no interactions between solvent and solute, the evaporation rate of the solvent would depend on the molecular weight of the solvent. Because of various interactions, this relationship is more complicated but solvent molecular weight does play an essential role in solvent diffusion. This is illustrated best by membranes which have pores sizes which limit the size of molecules which may pass through. The resistance of a material to solvents will be partially controlled by the molecular weight of the solvent since solvent molecules have to migrate to the location of the interactive material in order to interact with it. The chemical potential of a solvent also depends on its molecular weight (see eq. [6.6]). If all other influences and properties are equal, the solvent having the lower molecular weight is more efficiently dissolving materials, readily forms gels, and swells materials. All this is controlled by the molecular interactions between solvent and solute. In other words, at least one molecule of solvent involved must be available to interact with a particular segment of solute, gel, or network. If solvent molecular weight is low more molecules per unit weight are available to affect such changes. Molecular surface area and molecular volume are part of various theoretical estimations of solvent properties and they are in part dependent on the molecular mass of the sol25 vent. Many physical properties of solvents depend on their molecular weight, such as 20 boiling and freezing points, density, heat of evaporation, flash point, and viscosity. The relationship between these properties and 15 molecular weight for a large number of solvents of different chemical composition is affected by numerous other influences but 10 within the same chemical group (or similar structure) molecular weight of solvent correlates well with its physical properties. 5 Figure 2.3.1 gives an example of in0 500 1000 1500 2000 2500 terrelation of seemingly unrelated parameHeat of vaporization, kJ kg-1 ters: Hildebrand solubility parameter and Figure 2.3.1. Hildebrand solubility parameter vs. heat of heat of vaporization (see more on the subvaporization of selected solvents.

44

George Wypych

ject in the Section 2.3.19). As heat of vaporization increases, the solubility parameter also increases. Molar volume is a rather speculative, theoretical term. It can be calculated from Avogadro’s number but it is temperature dependent. In addition, free volume is not taken into consideration. Molar volume can be expressed as molecular diameter but solvent molecules are rather non-spherical therefore diameter is often misrepresentation of the real dimension. It can be measured from the studies on interaction but results differ widely depending on the model used to interpret results. 2.3.2 BOILING AND FREEZING POINTS Boiling and freezing points are two basic properties of solvents often included in specifications. Based on their boiling points, solvents can be divided to low (below 100oC), medium (100-150oC) and high boiling solvents (above 150oC). The boiling point of liquid is frequently used to estimate the purity of the liquid. A similar approach is taken for solvents. Impurities cause the boiling point of solvents to increase but this increase is very small (in the order of 0.01oC per 0.01% impurity). Considering that the error of boiling point can be large, contaminated solvents may be undetected by boiling point measurement. If purity is important it should be evaluated by some other, more sensitive methods. The difference between boiling point and vapor condensation temperature is usually more sensitive to admixtures. If this difference is more than 0.1oC, the presence of admixtures can be suspected. The boiling point can also be used to evaluate interactions due to the association among molecules of solvents. For solvents with low association, Trouton’s rule, given by the following equation, is fulfilled: ∆S

o bp

=

o ∆Hbp

Tbp

= 88 J mol −1 K −1

[2.3.1]

where: ∆Sobp ∆Hobp Tbp

molar change of enthalpy molar change of entropy boiling point

If the enthalpy change is high it suggests that the solvent has a strong tendency to form associations. Boiling point depends on molecular weight but also on structure. It is generally lower for branched and cyclic solvents. Boiling and freezing points are important considerations for solvent storage. Solvents are frequently stored under nitrogen blanket and they contribute to substantial emissions during storage. Freezing point of some solvents is above temperatures encountered in temperate climatic conditions. Although, solvents are usually very stable in their undercooled state, they rapidly crystallize when subjected to any mechanical or sonar impact. Figures 2.3.2 to 2.3.6 illustrate how the boiling points of individual solvents in a group are related to other properties. Figure 2.3.2 shows that chemical structure of a solvent affects the relationship between its viscosity and the boiling point. Alcohols, in particular, show a much larger change in viscosity relative to boiling point than do aromatic hydrocarbons, esters and ketones. This is caused by strong associations between molecules of alcohols, which contain hydroxyl groups. Figure 2.3.3 shows that alcohols are also less volatile


45

12 8

10 alcohols aromatic CH esters ketones

Viscosity, cP

6 5 4 3 2

Evaporation rate

7

alcohols aromatic CH esters ketones

8 6 4 2

1 0 50

100

150

200

250

300

0 50

Boiling point, oC

100

150

200

250

o

Boiling point, C Figure 2.3.2. Effect of boiling point on solvent viscosity.

Figure 2.3.3. Effect of boiling point on solvent evaporation rate (relative to butyl acetate = 1).

200

80 o


20

50

Flash point, C

60

0


150

-1

Specific heat, cal k mol

-1

100

100

150

200

250

o

Boiling point, C Figure 2.3.4. Specific heat of solvents vs. their boiling point.

100 50 0 -50 50

100

150

200

250

300

o

Boiling point, C Figure 2.3.5. Flash point of solvents vs. their boiling point.

than other three groups of solvents and for the same reason. Viscosity and evaporation rate of aromatic hydrocarbons, esters, and ketones follow single relationship for all three groups of solvents, meaning that the boiling point has strong influence on these two properties. There are individual points on this set of graphs which do not fall close to the fitted curves. These discrepancies illustrate that chemical interactions influence viscosity and evaporation rate. However, for most members of the four groups of solvents, properties correlate most strongly with boiling point. All linear relationships in Figure 2.3.4 indicate that specific heat is strongly related to the boiling point which is in agreement with the fact that boiling point is influenced by molecular weight. However, there are substantial differences in the relationships between dif-

46

George Wypych

30


15 10 5 0 50

100

o

20


150 Freezing point, C

Odor threshold, ppm

25

50 0 -50 -100

100

150

200

250

o

Boiling point, C Figure 2.3.6. Odor threshold of solvent vs. its boiling point.

-150 50

100

150

200

250

300

o

Boiling point, C Figure 2.3.7. Relationship between boiling and freezing points of solvents.

ferent groups of solvents and many experimental points are scattered. Figure 2.3.5 verifies the origin of flash point which has strong correlation with boiling point. Here, all four chemical groups of solvents have the same relationship. Odor threshold is an approximate but quite unreliable method of detection of solvent vapors. As the boiling point increases, the odor threshold (concentration in air when odor becomes detectable) decreases (Figure 2.3.6). This may suggest that slower evaporating solvents have longer residence time close to the source of contamination. Figure 2.3.7 shows the relationship between boiling and freezing points. A general rule is that the difference between boiling and freezing points for analyzed solvents is 190±30oC. Relatively small fraction of solvents does not follow this rule. Natural solvent mixtures such as aromatic or aliphatic hydrocarbons deviate from the rule (note that hydrocarbons in Figure 2.3.7 depart from the general relationship). If more groups of solvents is investigated, it will be seen that CFCs, amines, and acids tend to have a lower temperature difference between boiling and freezing points whereas some aliphatic hydrocarbons and glycol ethers have a tendency towards a larger difference. 2.3.3 SPECIFIC GRAVITY The specific gravity of most solvents is lower than that of water. When solvent is selected for extraction it is generally easy to find one which will float on the surface of water. Two groups of solvents: halogenated solvents and polyhydric alcohols have specific gravity greater than that of water. The specific gravity of alcohols and ketones increases with increasing molecular weight whereas the specific gravity of esters and glycol ethers decreases as their molecular weight decreases. The specific gravity of solvents affects their industrial use in several ways. Solvents with a lower density are more economical to use because solvents are purchased by weight but many final products are sold by volume. The specific gravity of solvent should be considered in the designs for storage systems and packaging. When switching the solvent types in storage tanks one must determine the weight of new solvent which can be accommodated in the tank. A container of CFC with a specific gravity twice that of most solvents, may be


47

1.44 1.55

1.42

1.45 1.4 alcohols aromatic CH esters ketones

1.35 1.3

0

50

100

150

200

250

Molecular weight, Daltons

Refractive index

Refreactive index

1.5

1.4 1.38 1.36 normal branched

1.34 1.32

1

2

3

4

5

6

7

8

Number of carbon atoms in alcohol Figure 2.3.8. Refractive index for four groups of solvents as the function of their molecular weight.

Figure 2.3.9. Refractive index of normal and branched alcohols as the function of number of carbon atoms in alcohol.

too heavy to handle. When metering by volume, temperature correction should always be used because solvent specific gravity changes substantially with temperature. 2.3.4 REFRACTIVE INDEX Refractive index is the ratio of the velocity of light of a specified wavelength in air to its velocity in the examined substance. When the principle of measurement is used it may be defined as the sine of the angle of incidence divided by the the sine of the angle of refraction. The absolute angle of refraction (relative to vacuum) is obtained by dividing the refractive index relative to air by a factor of 1.00027 which is the absolute refractive index of air. The ratio of the sines of the incident and refractive angles of light in the tested liquid is equal to the ratio of light velocity to the velocity of light in vacuum (that is why both definitions are correct). This equality is also referred to as Snell’s law. Figures 2.3.8 and 2.3.9 show the relationship between the molecular weight of a solvent and its refractive index. Figure 2.3.8 shows that there is a general tendency for the refractive index to increase as the molecular weight of the solvent increases. The data also indicates that there must be an additional factor governing refractive index. The chemical structure of the molecule also influences refractive index (Figure 2.3.9). Normal alcohols have a slightly higher refractive index than do branched alcohols. Cyclic alcohols have higher refractive indices than the linear and branched alcohols. For example, 1-hexanol has refractive index of of 1.416, 4-methyl-2-pentanol 1.41, and cyclohexanol 1.465. Aromatic hydrocarbons are not dependent on molecular weight but rather on the position of substituent in the benzene ring (e.g., m-xylene has refractive index of 1.495, o-xylene 1.503, and p-xylene 1.493). The data also show that the differences in refractive indices are rather small. This imposes restrictions on the precision of their determination. Major errors stem from poor instrument preparation and calibration and inadequate temperature control. The refractive index may change on average by 0.0005/oC. Refractive index is useful tool for determination of solvent purity but the precision of this estimation depends on relative difference be-

48

George Wypych

tween the solvent and the impurity. If this difference is small, the impurities, present in small quantities, will have little influence on the reading. 2.3.5 VAPOR DENSITY AND PRESSURE The relative vapor density of solvents is given by the following equation: d vp =

Ms M air

[2.3.2]

where: Ms Mair

the molecular mass of solvents the molecular mass of air (28.95 Daltons)

Figure 2.3.10 shows that the vapor density has linear correlation with molecular mass and that for both alcohols and ketones (as well as the other solvents) the relationship is similar. The data also show that solvent vapor densities are higher than air density. This makes ventilation a key factor in the removal of these vapors in the case of spill or emissions from equipment. Otherwise, the heavier than air vapors will flow along floors and depressions filling pits and subfloor rooms and leading to toxic exposure and/or risk of ignition and subsequent explosions. The Clausius-Clapeyron equation gives the relationship between molecular weight of solvent and its vapor pressure: d ln p MΛ = dT RT 2

[2.3.3]

where: p T M Λ R

vapor pressure temperature molecular mass of solvent heat of vaporization gas constant

Figure 2.3.11 shows that the vapor pressure of alcohols increases as the number of carbon atoms in the molecules and the molecular mass increases. A small increase in vapor pressure is produced when branched alcohols replace normal alcohols. Vapor pressure at any given temperature can be estimated by the use of Antoine (eq. [2.3.4]) or Cox (eq. [2.3.5]) equation (or chart). Both equations are derived from Clausius-Clapeyron equation: log p = A −

B C +T

[2.3.4]

log p = A −

B T + 230

[2.3.5]

where: A, B, C

constants. The constants A and B are different in each equation. The equations coincide when C = 230 in the Antoine equation.

From the above equations it is obvious that vapor pressure increases with temperature.


49

140 alcohols ketones

120 normal branched

100

o

5

Vapor pressure at 25 C, mm hg

Vapor density relative to air

6

4 3 2 1 20

40

60

60 40 20 0

80 100 120 140 160

Molecular weight, Daltons Figure 2.3.10. Vapor density relative to air of alcohols and ketones vs. their molecular weight.

80

1

2

3

4

5

6

7

8

Number of carbon atoms in alcohol Figure 2.3.11. Vapor pressure at 25oC of normal and branched alcohols vs. number of carbon atoms in their molecule.

The vapor pressure at the boiling point of a pure solvent is equal to atmospheric pressure. When solvents are used in mixtures or solutions, the vapor pressure is affected by other components present in the mixture. For example, if a solvent is hydrophilic, the addition of a hydrophilic solute decreases the vapor pressure. The addition of a hydrophobic solute to a hydrophilic solvent increases the vapor pressure. Alcohols have hydrophobic chains, therefore addition of small quantities of alcohol to water increases vapor pressure of resulting solution. Because of these phenomena and other types of associations between solvents in their mixtures, theory cannot be used to accurately predict the resulting vapor pressure. Raoult’s Law has limited prediction capability of the vapor pressure of two miscible solvents: p12 = m1 p1 + (1 − m1 )p 2

[2.3.6]

where: p12 m1 p1, p2

vapor pressure of the mixture molar fraction of the first component vapor pressures of the components

If associations exist between molecules in the mixture, the vapor pressure of the mixture is lower than that predicted by the law. 2.3.6 SOLVENT VOLATILITY The evaporation rate of solvents is important in many applications. This has resulted in attempts to model and predict solvent volatility. The evaporation rate of a solvent depends on its vapor pressure at the processing temperature, the boiling point, specific heat, enthalpy and heat of vaporization of the solvent, the rate of heat supply, the degree of association between solvent molecules and between solvent and solute molecules, the surface tension of the liquid, the rate of air movement above the liquid surface, and humidity of air surrounding the liquid surface.

50

George Wypych

The vapor pressure of solvent was found in the previous section to depend on its molecular weight and temperature. Figure 2.3.3 shows that the evaporation rate of a solvent may be predicted based on knowledge of its boiling point and Figure 2.3.4 shows that the specific heat of solvent also relates to its boiling point. The boiling point of solvent also depends on its molecular weight as does enthalpy and heat of vaporization. But there is not a high degree of correlation among these quantities because molecular associations exist which cannot be expressed by universal relationship. For this reason experimental values are used to compare properties of different solvents. The two most frequently used reference solvents are: diethyl ether (Europe) and butyl acetate (USA). The evaporation rate of other solvents is determined under identical conditions and the solvents are ranked accordingly. If diethyl ether is used as a reference point, solvents are grouped into four groups: high volatility < 10, moderate volatility 10-35, low volatility 35-50, and very low volatility > 50. If butyl acetate is used as the reference solvent, the solvents are grouped into three classes: rapid evaporation solvents > 3, moderate 0.8-3, and slow evaporating solvents < 0.8. In some applications such as coatings, casting, etc., evaporation rate is not the only important parameter. The composition must be adjusted to control rheological properties, prevent shrinkage, precipitation, formation of haze, and to provide the required morphology. Solvents with different evaporation rates can address all existing requirements. Both the surface tension of mixture and solvent diffusion affect the evaporation rate. This becomes a complex function dependent not only on the solvents present but also on influence of solutes on both surface tension and diffusion. These relationships affect the real evaporation rates of solvents from the complicated mixtures in the final products. In addition, solvent evaporation also depends on relative humidity and air movement. 2.3.7 FLASH POINT Flash point is the lowest temperature, corrected to normal atmospheric pressure (101.3 kPa), at which the application of an ignition source causes the vapors of a specimen to ignite under the specific conditions of the test. Flash point determination methods are designed to be applied to a pure liquid but, in practice, mixtures are also evaluated. It is important to understand limitations of such data. The flash point of a solvent mixture can be changed by adding various quantities of other solvents. For example, the addition of water or halogenated hydrocarbons will generally increase the flash point temperature of mixture. The flash point can also be changed by forming an azeotropic mixture of solvents or by increasing the interaction between solvents. At the same time, the flash point of single component within the mixture is not changed. If conditions during production, application, or in a spill allow the separation or removal of a material added to increase the flash point, then the flash point will revert to that of the lowest boiling flammable component. An approximate flash point can be estimated from the boiling point of solvent using the following equation: Flash point = 0.74Tb

[2.3.7]

Figure 2.3.5 shows that there is often good correlation between the two but there are instances where the relationship does not hold. The correlation for different groups of solvents varies between 0.89 to 0.96. Flash point can also be estimated from vapor pressure using the following equation:


51

100

o

Flash point, C

80

Lower flammability limit, %

7 alcohols ketones

60 40 20 0 -20

0

50

100

150

200

250

Vapor pressure, mm Hg Figure 2.3.12. Flash point vs. vapor pressure of solvent.

Flash point = a log p + b


6 5 4 3 2 1 0 -20

0

20

40

60

80 100 120 o

Flash point, C Figure 2.3.13. Lower flammability limit of solvents vs their flash points.

[2.3.8]

The constants a and b are specific to each group of solvents. Figure 2.3.12 shows that estimation of flash point from vapor pressure of solvent is less accurate than its estimation from boiling point. 2.3.8 FLAMMABILITY LIMITS Two limits of solvent flammability exist. The lower flammability limit is the minimum concentration of solvent vapor in oxidizing gas (air) that is capable of propagating a flame through a homogeneous mixture of the oxidizer and the solvent vapor. Below the lower flammability limit the mixture is too lean to burn or explode. The upper flammability limit is the maximum concentration of solvent vapor in an oxidizing gas (air) above which propagation of flame does not occur. Mixtures with solvent vapor concentrations above the upper flammability limit are too rich in solvent or too lean in oxidizer to burn or explode. The flammable limits depend on oxygen concentration, concentration of gases other than oxygen, the inert gas type and concentration, the size of the equipment, the direction of flame propagation, and the pressure, temperature, turbulence and composition of the mixture. The addition of inert gases to the atmosphere containing solvent is frequently used to reduce the probability of an explosion. It is generally assumed that if the concentration of oxygen is below 3%, no ignition will occur. The type of inert gas is also important. Carbon dioxide is more efficient inert gas than nitrogen. The size of equipment matters because of the uniformity of vapor concentration. A larger head space tends to increase the risk of inhomogeneity. The cooling effect of the equipment walls influences the evaporation rate and the vapor temperature and should be used in risk assessment. The flash point is not the temperature at which the vapor pressure in air equals the lower flammable limit. Although both parameters have some correspondence there are large differences between groups of solvents. There is a general tendency for solvents with a lower flammability limit to have a lower flash point. The flash point determination uses a

52

George Wypych

downward and horizontal propagation of flame. Flame propagation in these directions generally requires a higher vapor concentration than it is required for the upward flame propagation used to determine flammability limits. The flame in flash point determination is at some distance from the surface where the vapor concentration is at its highest (because vapors have higher density than air) than exists on the liquid surface thus flush analysis underestimates concentration of vapor. An increased vapor pressure typically increases the upper limit of flammability and reduces the lower limit of flammability. Pressures below atmospheric have little influence on flammability limits. An increase in temperature increases the evaporation rate and thus decreases the lower limit of flammability. There are a few general rules which help in the estimation of flammability limits. In the case of hydrocarbons, the lower limit can be estimated from simple formula: 6/number of carbon atoms in molecule; for benzene and its derivatives the formula changes to: 8/number of carbon atoms. To calculate the upper limits, the number of hydrogen and carbon atoms is used in calculation. The lower flammability limit of a mixture can be estimated from Le Chatelier’s Law: LFL mix =

100 φ1 φ φ + 2 +L+ n LFL1 LFL 2 LFL n

[2.3.9]

where: φi LFLi

fraction of components 1, 2, ..., n lower flammability limit of component 1, 2, ..., n

2.3.9 SOURCES OF IGNITION AND AUTOIGNITION TEMPERATURE Sources of ignition can be divided to mechanical sources (impact, abrasive friction, bearings, misaligned machine parts, choking or jamming of material, drilling and other maintenance operations, etc), electrical (broken light, cable break, electric motor, switch gear, liquid velocity, surface or personal charge, rubbing of different materials, liquid spraying or jetting, lightning, stray currents, radio frequency), thermal (hot surface, smoking, hot transfer lines, electric lamps, metal welding, oxidation and chemical reactions, pilot light, arson, change of pressure, etc.), and chemical (peroxides, polymerization, catalysts, lack of inhibitor, heat of crystallization, thermite reaction, unstable substances, decomposition reactions). This long list shows that when making efforts to eliminate ignition sources, it is also essential to operate at safe concentrations of volatile, flammable materials because of numerous and highly varied sources of ignition. The energy required for ignition is determined by the chemical structure of the solvent, the composition of the flammable mixture, and temperature. The energy of ignition of hydrocarbons decreases in the order alkanes > alkenes > alkynes (the presence of double or triple bond decreases the energy energy required for ignition). The energy requirement increases with an increase in molecular mass and an increase branching. Conjugated structure generally requires less ignition energy. Substituents increase the required ignition energy in the following order: mercaptan < hydroxyl < chloride < amine. Ethers and ketones require higher ignition energy but an aromatic group has little influence. Peroxides require extremely little energy to ignite.

53

1.4

Minimum ignition energy, mJ

Minimum spark ignition energy, mJ


1

0.6

0.2 0.9

1.1

1.3

1.5

1.7

Stoichiometric air to solvent ratio Figure 2.3.14. Minimum ignition energy vs. stoichiometric ratio of air to methyl ethyl ketone. [Data from H F Calcote, C A Gregory, C M Barnett, R B Gilmer, Ind. Eng. Chem., 44, 2656 (1952)].


0.4

ethyl acetate toluene ethanol

0.3

0.2

0.1 20

60

100

140

Temperature, oC Figure 2.3.15. Minimum ignition energy vs. temperature for selected solvents. [Data from V S Kravchenko, V A Bondar, Explosion Safety of electrical Discharges and Frictional Sparks, Khimia, Moscow, 1976].

o

Autoignition temperature, C

Figure 2.3.14 shows the effect of changing the ratio of air to methyl ethyl 500 ketone on the minimum spark ignition energy. The ignition energy decreases within 450 the studied range as the amount of air increases (less flammable content). Figure 400 2.3.15 shows the effect of temperature on the minimum ignition energy of selected 350 solvents. There are differences between solvents resulting from differences in 300 chemical structure as discussed above but the trend is consistent − a decrease of re250 quired energy as temperature increases. 20 40 60 80 100 120 140 160 The autoignition temperature is the Molecular weight, Daltons minimum temperature required to initiate Figure 2.3.16. Autoignition temperature of selected combustion in the absence of a spark or solvents vs. their molecular weight. flame. The autoignition temperature depends on the chemical structure of solvent, the composition of the vapor/ air mixture, the oxygen concentration, the shape and size of the combustion chamber, the rate and duration of heating, and on catalytic effects. Figure 2.3.16 shows the effect of chemical structure on autoignition temperature. The general trend for all groups of solvents is that the autoignition temperature decreases as molecular weight increases. Esters and ketones behave almost identically in this respect and aromatic hydrocarbons are very similar. The presence of a hydroxyl group substantially reduces autoignition temperature.

54

George Wypych

6000

Heat of combustion, kJ mol

-1

The effect of the air to solvent ratio on autoignition temperature is similar to that 5000 on ignition energy (see Figure 2.3.14). As the oxygen concentration increases within 4000 the range of the flammability limits, the autoignition temperature increases. The 3000 autoignition temperature increases when the size of combustion chamber decreases. Rapid heating reduces the autoignition 2000 temperature and catalytic substance may drastically reduce it. 1000 2.3.10 HEAT OF COMBUSTION 0 (CALORIFIC VALUE) 1 2 3 4 5 6 7 8 Heat of combustion, also known as caloNumber of carbon atoms rific value, is the quantity of energy per Figure 2.3.17. Heat of combustion vs. number of carbon mole released during combustion. It coinatoms in molecule. cides with the heat of reaction. Solvents have higher heats of combustion than typical fuels such as natural gas, propane or butane. They can be very good source of energy in plants which process solutions. In addition to supplying energy, the combustion of solvents can be developed to be one of the cleanest method of processing from solutions. Two approaches are commonly used: solvent vapors are directed to a combustion chamber or spent solvents are burned in furnaces. The heat of combustion of a liquid solvent is less than 1% lower than the heat of combustion of a vapor. Figure 2.3.17 shows the relationship between the heat of combustion and the number number of carbon atoms in the molecule. The heat of combustion increases as molecular weight increases and decreases when functional groups are present. 2.3.11 HEAT OF FUSION Heat of fusion is the amount of heat to melt the frozen solvent. It can be used to determine the freezing point depression of solute. 2.3.12 ELECTRIC CONDUCTIVITY Electric conductivity is the reciprocal of specific resistance. The units typically used are either ohm-1 m-1 or, because the conductivities of solvents are very small picosiemens per meter which is equivalent to10-12 ohm-1 m-1. The electric conductivity of solvents is very low (typically between 10-3 - 10-9 ohm-1 m-1). The presence of acids, bases, salts, and dissolved carbon dioxide might contribute to increased conductivity. Free ions are solely responsible for the electric conductivity of solution. This can be conveniently determined by measuring the conductivity of the solvent or the conductivity of the water extract of solvent impurities. The electronic industry and aviation industry have the major interest in these determinations. 2.3.13 DIELECTRIC CONSTANT (RELATIVE PERMITTIVITY) The dielectric constant (or relative permittivity) of a solvent reflects its molecular symmetry. The value of the dielectric constant is established from a measurement relative to vacuum. The effect is produced by the orientation of dipoles along an externally applied n-hydrocarbons alcohols


55

35

35

Dielectric constant

30 25 20 15 10 5

alcohols aliphatic CH esters ketones

30 Dielectric constant

alcohols aliphatic CH esters ketones

25 20 15 10 5

0 20

40

60

80

100 120 140

Molecular weight, Daltons Figure 2.3.18. Dielectric constant vs. molecular weight of selected solvents.

0 1.3

1.35

1.4

1.45

1.5

Refractive index Figure 2.3.19. Dielectric constant of selected solvents vs. their refractive index.

electric field and from the separation of charges in apolar molecules. This orientation causes polarization of the molecules and a drop in electric field strength. Dielectric constant data may be used in many ways. In particular, it is the factor which permits the evaluation of electrostatic hazards. The rate of charge decay is a product of dielectric constant and resistivity. In solvent research, dielectric constant has a special place as a parameter characteristic of solvent polarity. The dielectric constant, ε, is used to calculate dipole moment, µ: µ=

ε −1 VM ε−2

[2.3.10]

where: VM

molar volume

The product of dipole moment and dielectric constant is called the electrostatic factor and it is a means of classifying solvents according to their polarity. Figure 2.3.18 shows that the dielectric constant correlates with molecular weight. It is only with aliphatic hydrocarbons that the dielectric constant increases slightly as the molecular weight increases. The dielectric constant of alcohols, esters, and ketones decreases as their molecular weight increases, but only alcohols and ketones have the same relationship. The dielectric constants of esters are well below those of alcohols and ketones. The dielectric constant also correlates with refractive index. In the case of aliphatic hydrocarbons, the dielectric constant increases slightly as refractive index increases. Both aromatic and aliphatic hydrocarbons have dielectric constants which follow the relationship: ε ≈ n D2 . The dielectric constants of alcohols, esters and ketones decrease as the refractive constants increase but only alcohols and ketones form a similar relationship. The dielectric constants of ketones poorly correlate with their refractive indices.

56

George Wypych

2.3.14 OCCUPATIONAL EXPOSURE INDICATORS The measurement of solvent concentration in the workplace place is required by national regulations. These regulations specify, for individual solvents, at least three different concentrations points: the maximum allowable concentration for an 8 hour day exposure, the maximum concentration for short exposure (either 15 or 30 min), and concentration which must not be exceeded at any time. These are listed in the regulations for solvents. The listing is frequently reviewed and updated by the authorities based on the most currently available information. In the USA, the threshold limit value, time-weighted average concentration, TLV-TWA, is specified by several bodies, including the American Conference of Governmental Industrial Hygienists, ACGIH, the National Institute of Safety and Health, NIOSH, and the Occupational Safety and Health Administration, OSHA. The values for individual solvents stated on these three lists are very similar. Usually the NIOSH TLV-TWA are lower than on the other two lists. Similar specifications are available in other countries (for example, OES in UK, or MAK in Germany). The values for individual solvents are selected based on the presumption that the maximum allowable concentration should not cause injury to a person working under these conditions for 8 hours a day. For solvent mixtures, the following equation is used in Germany to calculate allowable limit: ci i =1 MAK i

i =n

I MAK = ∑

[2.3.11]

where: IMAK c MAK

evaluation index concentrations of components 1, 2, ..., n maximum permissible concentrations for components 1, 2, ..., n

The maximum concentrations for short exposure is the most frequently limit specified for an exposure of 15 min with a maximum of 4 such occurrences per day each occurring at least 60 min apart from each other. These values are 0-4 times larger than TLVs. They are selected based on the risks associated with an individual solvent. Solvent concentrations which should not be exceeded at any time are seldom specified in regulations but, if they are, the values stated as limits are similar to those on the three lists. In addition, to maintaining concentration below limiting values, adequate protection should be used to prevent the inhalation of vapors and contact with the skin (see Chapter 24). 2.3.15 ODOR THRESHOLD The principal for odor threshold was developed to relate the human sense of smell to the concentration of the offending substances. If the substance is toxic, its detection may provide early warning to the danger. However, if the odor threshold is higher than the concentration at which harm may be caused it is not an effective warning system. Toxic substances may have very little or no odor (e.g., carbon monoxide) and an individual’s sense of smell may vary widely in its detection capabilities. A knowledge of odor threshold is most useful in determining the relative nuisance factor for an air pollutant when designing a control system to avoid complaints from neighboring people surrounding a facility. Regulations often state (as they do for example in Ontario, Canada) that, even when the established concentra-


57

tion limits for air pollutants are met, if neighbors complain, penalties will be applied. Figure 2.3.6 shows that odor threshold is related to the boiling point (although odor threshold decreases with boiling point increasing). It is known from comparisons of TLV and odor detection that odor detection is not a reliable factor. 2.3.16 TOXICITY INDICATORS Lethal dose, LD50, and lethal concentration, LC50, are commonly used indicators of substance toxicity. LD50 is reported in milligrams of substance per kilogram of body weight to cause death in 50% of tested animals (exception is LC50 which is given in ppm over usually the period of 4 hours to produce the same effect). It is customary to use three values: LD50-oral, LD50-dermal, and LC50-inhalation which determine the effect of a chemical substance on ingestion, contact with the skin and inhalation. The preferred test animal for LD50-oral and LC50-inhalation is the rat. The rabbit is commonly used for LD50-dermal determination but other test animals are also used. There is no official guideline on how to use this data but the Hodge-Sterner table is frequently referred to in order to assign a particular substance to a group which falls within certain limits of toxicity. According to this table, dangerously toxic substances are those which have LD50 < 1 mg/kg, seriously toxic - 1-50, highly toxic - 50-500, moderately toxic 500-5,000, slightly toxic - 5,000-15,000, and extremely low toxic - >15,000 mg/kg. Using this classification one may assess the degree of toxicity of solvents based on a lethal dose scale. No solvent is classified as a dangerously toxic material. Ethylenediaminetetraacetic acid and furfural are seriously toxic materials. Butoxyethanol, ethylene oxide, formaldehyde, metasulfonic acid, 3-methyl-2-butanone, N-nitrosodimethylamine, and triethylamine are classified as highly toxic material. The remaining solvents fall into the moderately, slightly, and extremely low toxic material classes. The LD50-oral is usually assigned a lower value than LD50-dermal but there are many cases where the opposite applies. Toxicity information is usually further expanded by adding more details regarding test animals and target organs. In addition to estimates of toxicity for individual solvents, there are lists which designate individual solvents as carcinogenic, mutagenic, and reproductively toxic. These lists contain the name of solvent with yes or no remark (or similar). If a solvent is not present on the list that does not endorse its benign nature because only materials that have been tested are included in the lists. To further elaborate, materials are usually divided into three categories: substance known to cause effect on humans, substance which has caused responses in animal testing and given reasons to believe that similar reactions can be expected with human exposures, and substance which is suspected to cause responses based on experimental evidence. In the USA, four agencies generate lists of carcinogens. These are: the Environmental Protection Agency, EPA, the International Agency for Research on Cancer, IRAC, the National Toxicology Program, NTP, and the Occupational Safety and Health Administration, OSHA. Although, there is a good agreement between all four lists, each assessment differs in some responds. The following solvents made at least one of the lists (no distinction is given here to the category assignment but any known or suspected carcinogen found on any list is given (for more details see Chapter 3)): acetone, acrolein, benzene, carbon tetrachloride, dichloromethane, 1,4-dioxane, ethylene oxide, formaldehyde, furfural, d-limonene, N-nitrosodimethyl amine, propylene oxide, tetrachloroethylene, 2,4-toluenediisocyanate, 1,1,2-trichloroethylene, and trichloromethane.

58

George Wypych

Mutagenic substances have the ability to induce genetic changes in DNA. The mutagenicity list maintained in the USA includes the following solvents: all solvents listed above for carcinogenic properties with exception of dichloromethane, d-limonene, and tetrachloroethylene. In addition, the following long list solvents: 1-butanol, 2-butanol, γ-butyrolactone, 2-(2-n-butoxyethoxy)ethanol, chlororodifluoromethane, chloromethane, diacetone alcohol, dichloromethane, diethyl ether, dimethyl amine, dimethylene glycol dimethyl ether, dimethyl sulfoxide, ethanol, 2- ethoxyethanol, 2-ethoxyethanol acetate, ethyl acetate, ethyl propionate, ethylbenzene, ethylene glycol diethyl ether, ethylene glycol methyl ether acetate, ethylene glycol monophenyl ether, ethylenediaminetetraacetic acid, formic acid, furfuryl alcohol, heptane, hexane, methyl acetate, 3-methyl-2-butanol, methyl ester of butyric acid, methyl propionate, N-methylpyrrolidone, monomethylamine, 1-octanol, 1-pentanol, 1-propanol, propyl acetate, sulfolane, 1,1,1-trichloroethane, triethylene glycol, triethylene glycol dimethyl ether, trifluoromethane, trimethylene glycol, and xylene (mixture only). It is apparent that this much longer list includes commonly used solvents from the groups of alcohols, halogenated solvents, hydrocarbons, glycols, and esters. The following solvents are reported to impair fertility: chloroform, ethylene glycol and its acetate, 2-methoxypropanol, 2-methoxypropyl acetate, dichloromethane, methylene glycol and its acetate, and N,N-dimethylformamide. 2.3.17 OZONE - DEPLETION AND CREATION POTENTIAL Ozone depletion potential is measured relative to CFC-11 and it represents the amount of ozone destroyed by emission of a vapor over its entire atmospheric lifetime relative to that caused by the emission of the same mass of CFC-11. Urban ozone formation potential is expressed relative to ethene. It represents the potential of an organic solvents vapor to form ozone relative to that of ethene ((g O3/g solvent)/(g O3/g ethene)). Several groups of solvents, including alcohols, aldehydes, amines, aliphatic and aromatic hydrocarbons, esters, ethers, and ketones are active in ozone formation. Aldehydes, xylenes, some unsaturated compounds, and some terpenes are the most active among those. 2.3.18 OXYGEN DEMAND There are several indicators of solvent biodegradation. Most solvents have a biodegradation half-life of days to weeks and some biodegrade even faster. The amount of oxygen required for its biodegradation is a measure of a solvent’s impact on natural resources. Several factors are used to estimate this, such as biological oxygen demand, BOD, after 5-day and 20-day aerobic tests, chemical oxygen demand, COD, and theoretical oxygen demand, TOD. All results are given in grams of oxygen per gram of solvent. COD is the amount of oxygen removed during oxidation in the presence of permanganate or dichromate. TOD is the theoretically calculated amount of oxygen required to oxidize solvent to CO2 and H2O. Most alcohols and aromatic hydrocarbons have a highest BOD5. They consume twice their own weight in oxygen. 2.3.19 SOLUBILITY The prediction of solubility of various solutes in various solvents is a major focus of research. An early theory has been that “like dissolves like”. Regardless of the apparent merits of this theory it is not sufficiently rigorous and is overly simple.


59

A universal approach was developed by Hildebrand who assumed that the mutual solubility of components depends on the cohesive pressure, c. The square root of cohesive pressure is the Hildebrand’s solubility parameter, δ: δ= c =

∆Hv − RT Vm

[2.3.12]

where: ∆Hv R T Vm M d

heat of vaporization gas constant temperature molar volume of solvent = M/d molecular mass of solvent density of solvent

Frequently, the term RT is neglected because it accounts for only 5-10% of the heat of vaporization. This equation explains the reasons for the correlation between the Hildebrand solubility parameter and heat of vaporization as given in Figure 2.3.1. The Hildebrand model takes into account only the dimensions of molecules or of the molecular segments participating in the process of solvation and dispersion interactions. The model is useful, therefore, in predicting the solubility of non-polar substrates. The solubility parameters of solvents and solutes are compared and if they are similar there is high probability (exceptions exist) that the solvents are miscible that a solute is soluble in a solvent. Two solvents having the same solubility parameters should have the same dissolving capabilities. If one solvent has solubility parameter slightly below the solubility parameter of solute and the second solvent has solubility parameter above the solute, the mixture of both solvents should give better results than either solvent alone. This model is an experimental and mathematical development of the simple rule of “like dissolves like”. Solvents and solutes also interact by donor-aceptor, electron pair, and hydrogen bonding interactions. It can be predicted that the above concept is not fully universal, especially in the case of solutes and solvents which may apply these interactions in their solubilizing action. Hansen developed a three-dimensional scale with parameters to expand theory in order to include these interactions. Hansen defined solubility parameter by the following equation: δ 2 = δ d2 + δ p2 + δ h2

[2.3.13]

where: δd δp δh

dispersion contribution to solubility parameter polar contribution to solubility parameter hydrogen bonding contribution to solubility parameter

Hansen defined solvent as a point in three-dimensional space and solutes as volumes (or spheres of solubility). If a solvent point is within the boundaries of a solute volume space then the solute can be dissolved by the solvent. If the point characterizing the solvent is outside the volume space of a solute (or resin) such a solvent does not dissolve the solute. The solubility model based on this concept is broadly applied today by modern computer techniques using data obtained for solvents (the three components of solubility parameters) and solutes (characteristic volumes). A triangular graph can be used to outline the limits of

60

George Wypych

55

140

Kauri butanol number

2

R = 0.994 2

R = 0.996

100

2

R = 0.992

80 60 aliphatic CH aromatic CH CFC

40

Kauri butanol number

50

120

45 40 35 30 2

20

y = 62.2 - 0.47x R = 0.996

7

7.5

8

8.5

9

9.5

10

Hildebrand solubility parameter, cal-1/2 cm-3/2

Figure 2.3.20. Kauri butanol number vs. Hildebrand solubility parameter.

25 10

20

30

40

50

60

70

80

o

Aniline point, C

Figure 2.3.21. Kauri butanol number vs. aniline point.

solubility and place different solvents within the matrix to determine their potential dissolving capability for a particular resin. Simpler methods are also used. In the paint industry, Kauri butanol values are determined by establishing the tolerance of a standard solution of Kauri resin in n-butanol to the addition of diluents. This method is applicable to hydrocarbons (both aromatic and aliphatic) and CFCs. Figure 2.3.20 shows that there is a good correlation between the Kauri butanol number and the Hildebrand solubility parameter. The Kauri butanol number can be as high as 1000 (amyl ester of lactic acid) or 500 (Freon solvent M-162). The aniline point determination is another method of establishing the solubilizing power of a solvent by simple means. Here, the temperature is measured at which a solution just becomes cloudy. Figure 2.3.21 shows that there is a good correlation between the Kauri number and the aniline point. Also dilution ratio of cellulose solution is measured by standardized methods (see Chapter 15). 2.3.20 OTHER TYPICAL SOLVENT PROPERTIES AND INDICATORS There are many other solvent properties and indices which assist in solvent identification and selection and help us to understand the performance characteristics of solvents. Most data characterizing the most important properties were discussed in the sections above. The solvent properties and classification indicators, which are discussed below, are included in the Solvent Database available on CD-ROM from ChemTec Publishing. Name. A solvent may have several names such as common name, Chemical Abstracts name, and name according to IUPAC systematic nomenclature. Common names have been used throughout this book and in the CD-ROM database because they are well understood by potential users. Also, CAS numbers are given in the database to allow user of the database to use the information with Chemical Abstract searches. In the case of commercial solvents which are proprietary mixtures, the commercial name is used. The molecular formula for each solvent is given in the database, followed by the molecular formula in Hills notation, and the molecular mass (if solvent is not a mixture).


61

-3/2

The CAS number identifies the chemical compound or composition without ambiguity. 14 RTECS number is the symbol given 13 by Registry of Toxic Effects of Chemical Substances (e.g. AH4025000) to identify 12 toxic substances. 11 Composition is given for solvents 10 which are manufactured under trade name and have a proprietary composition (if such 9 information is available). 8 Solvent purity (impurities) is given as a percent and known impurities and their 7 20 25 30 35 40 45 concentrations are provided. Hygroscopicity and water solubility Surface tension, dynes cm-2 Figure 2.3.22. Hildebrand solubility parameter vs. of solvents is an important characteristic in surface tension for four groups of solvents. many applications. The data are given in the database either in mg of water per kilogram of solvent or in a generic statement (e.g. miscible, slight). Many solvents are hygroscopic, especially those which contain hydroxyl groups. These solvents will absorb water from their surrounding until equilibrium is reached. The equilibrium concentration depends on the relative humidity of air and the temperature. If solvents must maintain a low concentration of water, vents of storage tanks should be fitted with silica gel or molecular sieves cartridges or tanks should be sealed and equipped with pressure and vacuum relief vents which open only to relieve pressure or to admit a dry inert gas to replace the volume pumped out. Preferably prevention of water from contacting solvents or the selection of solvents with a low water content is more economical than the expensive operation of drying a wet solvent. Surface tension and solubility parameter have been related in the following equation: alcohols aromatic CH esters ketones

1/2

Hildebrand solubility parameter, cal cm

15

 γ  δ = 21 . K  1/ 3  V 

a

[2.3.14]

where: δ K, a γ V

Hildebrand solubility parameter constants surface tension molar volume of the solvent

However, Figure 2.3.22 shows that the parameters correlate only for aromatic hydrocarbons. For three other groups the points are scattered. The equation has a very limited predictive value. Viscosity. Figure 2.3.2 shows that viscosity of solvents correlates with their boiling point. There are substantial differences in the viscosity - boiling point relationship among alcohols and other groups of solvents. These are due to the influence of hydrogen bonding on the viscosity of alcohols.

62

George Wypych

Thermal conductivity of solvents is an important property which determines the heat transfer in a solvent or solution and influences the evaporation rate of solvents as a solution is being dried. Activity coefficients may be applied to different processes. In one application, the activity coefficient is a measure of the escaping tendency from liquid to another liquid or a gaseous phase (in the liquid to gas phase they can be quantified using Henry’s Law coefficient). These activity coefficients are derived from distillation data at temperatures near the boiling point or from liquid-liquid extraction calculations. In another application as defined by Hildebrand and Scratchard solvent activity to dissolve a non-electrolyte solute is given by equation: lnf =

Vm (δsolute − δsovent )

2

RT

[2.3.15]

where: f Vm δsolute δsolvent RT

activity coefficient molar volume of solute solubility parameter of solute solubility parameter of solvent gas constant x temperature

This coefficient is used to express rate constants of bimolecular reactions. Azeotropes. One solvent may form azeotropes with another solvent due to molecular association. This physical principle can be exploited in several ways. The most important in solvent applications is the possibility of reducing the boiling temperature (some azeotropes have lower boiling point) therefore an applied product such as a coating may lose its solvents and dry faster. The formation of such azeotrope also lowers flash point by which it increases hazards in product use. The formation of an azeotrope is frequently used to remove water from a material or a solvent. It affects the results of a distillation since azeotrope formation makes it difficult to obtain pure components from a mixture by distillation. Azeotrope formation can be suppressed by lowering the boiling point (distillation under vacuum). One benefit of azeotropic distillation is the reduction in the heat required to evaporate solvents. Henry’s constant is a measure of the escaping tendency of a solvent from a very dilute solution. It is given by a simple equation: Henry’s constant = p × φ, where p is the pressure of pure solvent at the solution temperature and φ is the solvent concentration in the liquid phase. A high value of Henry’s constant indicates that solvent can be easily stripped from dilute water solution. It can also be used to calculate TLV levels by knowing concentration of a solvent in a solution according to the equation: TLV (in ppm) = [18 H (concentration of solvent in water)]/ molecular weight of solvent. pH and corrosivity. The pH of solvents is of limited value but it is sometimes useful if the solvent has strong basic or acidic properties which could cause corrosion problems. The acid dissociation constant is the equilibrium constant for ionization of an acid and is expressed in negative log units. The color of a solvent may influence the effect of solvent on the final product and allow the evaluation of solvent quality. The colorless solvents are most common but there are many examples of intrinsically colored solvents and solvents which are colored because of an admixture or inadequate storage conditions or too long storage.


63

Odor. Odor threshold (discussed above) is not a precise tool for estimation concentration of vapors. The description of odor has little relevance to the identification of solvent but description of odors in the database may be helpful in the selection of solvent to minimize odor or make it less intrusive. UV absorption maxima for different solvents given in the database are useful to predict the potential effect of solvent on UV absorption from sun light. The collection of data is also useful for analytical purposes. Solvent partition - activated carbon and between octanol and water. Solvents can be economically removed from dilute solutions by activated carbon or ion exchange resins. Activated carbon partition coefficient which helps to determine the amount of activated carbon needed to remove a contaminant can be obtained using the following equation: m=

w Pc

[2.3.16]

where: w P c

total weight of solvent in solution partition coefficient residual concentration of solvent remaining after treatment.

The octanol/water partition coefficient is the log of solubility of the solvent in water relative to that in octanol. This coefficient is used to estimate biological effects of solvents. It can also be used to estimate the potential usefulness of a solvent extraction from water by any third solvent. Soil adsorption constant is a log of the amount of a solvent absorbed per unit weight of organic carbon in soil or sediment. Atmospheric half-life of solvents due to reaction in the atmosphere with hydroxyl radicals and ozone is a measure of the persistence of particular solvent and its effect on atmospheric pollution. Hydroxyl rate constant is the reaction rate constant of the solvent with hydroxyl radicals in the atmosphere. Global warming potential of a well-mixed gas is defined as the time-integrated commitment to radiative forcing from the instantaneous release of 1 kg of trace gas expressed relative to that from the release of 1 kg of CO2. Biodegradation half-life determines persistence of the solvent in soil. Commercial proprietary solvents mixtures are classified as biodegradable and solvents having known chemical compositions are classified according to the time required for biodegradation to cut their initial mass to half. Target organs most likely affected organs by exposure to solvents. The database contains a list of organs targeted by individual solvents. Hazchem Code was developed in the UK for use by emergency services to determine appropriate actions when dealing with transportation emergencies. It is also a useful to apply as a label on storage tanks. It consists a number and one or two letters. The number informs about the firefighting medium to be used. The first letter gives information on explosion risk, personal protection and action. A second letter (E) may be added if evacuation is required.

3

Production Methods, Properties, and Main Applications 3.1 DEFINITIONS AND SOLVENT CLASSIFICATION George Wypych ChemTec Laboratories, Inc., Toronto, Canada

Several definitions are needed to classify solvents. These are included in Table 3.1.1 Table 3.1.1 Definitions Term

Definition

Solvent

A substance that dissolves other material(s) to form solution. Common solvents are liquid at room temperature but can be solid (ionic solvents) or gas (carbon dioxide). Solvents are differentiated from plasticizers by limiting their boiling point to a maximum of 250oC. To differentiate solvents from monomers and other reactive materials - a solvent is considered to be non-reactive.

Polarity

Polarity is the ability to form two opposite centers in the molecule. The concept is used in solvents to describe their dissolving capabilities or the interactive forces between solvent and solute. Because it depends on dipole moment, hydrogen bonding, entropy, and enthalpy, it is a composite property without a physical definition. The dipole moment has the greatest influence on polar properties of solvents. Highly symmetrical molecules (e.g. benzene) and aliphatic hydrocarbons (e.g. hexane) have no dipole moment and are considered non-polar. Dimethyl sulfoxide, ketones, esters, alcohol are examples of compounds having dipole moments (from high to medium, sequentially) and they are polar, medium polar, and dipolar liquids.

Polarizability

The molecules of some solvents are electrically neutral but dipoles can be induced by external electromagnetic field.

Normal

A normal solvent does not undergo chemical associations (e.g. the formation of complexes between its molecules).

66

George Wypych

Term

Definition

Aprotic/Protic

Aprotic solvents (also commonly called inert) have very little affinity for protons and are incapable to dissociating to give protons. Aprotic solvents are also called indifferent, non-dissociating, or non-ionizing. Protic solvents contain proton-donating groups.

Protogenic

An acidic solvent capable of donating protons.

Protophilic

A basic solvent able to combine with hydrogen ion or to act as a proton acceptor.

Acidic/Basic

Lewis acidity/basicity determines the solvent’s ability to donate or accept a pair of electrons to form a coordinate bond with solute and/or between solvent molecules. A scale for this acid/base property was proposed by Gutman (DN and AN donor and acceptor number, respectively) based on calorimetric determination. The complete proton transfer reaction with formation of protonated ions is determined by proton affinity, gas phase acidity, acid or base dissociation constants. Both concepts differ in terms of net chemical reaction.

Hydrogen-bonding

A bond involving a hydrogen atom, which is bound covalently with another atom, is referred to as hydrogen bonding. Two groups are involved: hydrogen donor (e.g., hydroxyl group) and hydrogen acceptor (e.g., carbonyl group).

Solvatochromism

Shift of UV/Vis absorption wavelength and intensity in the presence of solvents. A hypsochromic (blue) shift increases as solvent polarity increases. The shift in the red direction is called bathochromic.

Dielectric constant

A simple measure of solvent polarity (the electrostatic factor is a product of dielectric constant and dipole moment). The electrical conductivity of solvent indicates if there is a need to earth (or ground) the equipment which handles solvent to prevent static spark ignition. Admixtures affect solvent conductivity. These are most important in electronics industry.

Miscible

Solvents are usually miscible when their solubility parameters do not differ by more than 5 units. This general rule does not apply if one solvent is strongly polar.

Good solvent

Substances readily dissolve if the solubility parameters of solvent and solute are close (less than 6 units apart). This rule has some exceptions (for example, PVC is not soluble in toluene even though the difference of their solubility parameters is 2.5).

Θ solvent

The term relates to the temperature of any polymer/solvent pair at which chain expansion is exactly balanced by chain contraction. At this temperature, called Θ temperature chain dimensions are unperturbed by long-range interactions.

Reactivity

Solvent, according to this definition, should be a non-reactive medium but in some processes solvent will be consumed in the reaction to prevent its evaporation (and pollution). Solvents affect reactivity in two major ways: viscosity reduction and decreasing the barrier of Gibbs activation energy.

Hygroscopicity

Some solvents such as alcohols and glycols are hygroscopic and, as such, are unsuitable for certain applications which require a moisture-free environment or a predetermined freezing point. Solvents which are not hygroscopic may still contain moisture from dissolved water.

3.1 Definitions and solvent classification

Term

67

Definition

Solvent strength

Solvent strength is used to establish required solvent concentration to form a clear solution and to estimate the diluting capabilities of pre-designed system. Two determined quantities are used for the purpose: Kauri butanol value and aniline point.

Solvent partition

Solvent partition is determined for three purposes: to estimate the potential for solvent removal from dilute solution by carbon black adsorption, to evaluate the partition of solute between water and solvent for the purpose of studying biological effects of solvents and solutes, and to design system for solvent extraction.

Volatility

Solvent volatility helps in estimation of the solvent evaporation rate at temperatures below its boiling point. The Knudsen, Henry, Cox, Antoine, and Clausius-Clapeyron equations are used to estimate the vapor pressure of a solvent over a liquid, its evaporation rate, and the composition of the atmosphere over the solvent. The boiling point of a solvent gives an indication of its evaporation rate but it is insufficient for its accurate estimation because of the influence of the molar enthalpy of evaporation.

Residue

This may refer to either the non-volatile residue or the potential for residual solvent left after processing. The former can be estimated from the solvent specification, the later is determined by system and technology design.

Carcinogenic

Solvents may belong to a group of carcinogenic substances. Several groups of solvents have representatives in this category (see listings in Section 3.3)

Mutagenic

Mutagenic substance causes genetic alterations, such as genetic mutation or a change to the structure and number of chromosomes (mutagens listed in Section 3.3).

Impairing reproduction

Several solvents in the glycols and formamides groups are considered to impair fertility.

Toxicity

LD50 and LC50 give toxicity in mg per kg of body weight or ppm, respectively. Threshold limit values place a limit on permissible concentration of solvent vapors in the work place. Also “immediate-danger-to-life” and “short-term-exposure-limits” are specified for solvents. Odor threshold values have limited use in evaluating the potential danger to solvent exposure.

Flammable

Several data are used to evaluate the dangers of solvent explosion and flammability. Flash point and autoignition temperature are used to determine a solvent’s flammability and its potential for ignition. The flash points for hydrocarbons correlate with their initial boiling points. Lower and upper explosive limits determine the safe ranges of solvent concentration.

Combustible

The net heat of combustion and the calorific value help to estimate the potential energy which can be recovered from burning used solvents. In addition, the composition of the combustion products is considered to evaluate potential corrosiveness and the effect on the environment.

Ozone depleter

Ozone depletion potential is the value relative to that of CFC-11. It represents the amount of ozone destroyed by the emission of gas over its entire atmospheric life-time. Photochemical ozone creation potential is a relative value to that of ethene to form ozone in an urban environment. Numerous solvents belong to both groups.

68

George Wypych

Term

Definition

Biodegradability

Several methods are used to express biodegradability. These include biodegradation half-life, biological oxygen demand, chemical and theoretical oxygen demand.

Cost

Cost of solvent is a key factor in solvent selection.

The above list of terms and parameters is not exhaustive. These and related subjects are discussed at length in various parts of the book. The table is presented to assist in an understanding solvent classification. A review of the selected definitions suggests that there are many important determinants of solvent quality for specific application. Some solvent parameters are conflicting, some not well quantified, and each solvent application requires a unique set of solvent performance criteria. It can be thus anticipated, prior to any analysis, that the chemical structure can be used as the best means of solvent classification for any application. Such a classification is used in this book because of its broad application. Chemical names used are the common names because they are generally understood by all solvents users. Other means of classification are briefly analyzed below because they are useful in some applications. For a classification to be useful, it must be based on a model and a method which permits its quantification. In organic synthesis, the solvent’s polarity plays an important role. Dimroth and Reichardt1 developed a classification based on the normalized empirical parameter of solvent polarity, E NT , given by the following equation: E TN =

E T (solvent ) − E T (TMS ) E T (solvent) − 30.7 = 32.4 E T (water ) − E T (TMS )

[3.1.1]

where: ET TMS

excitation energy tetramethylsilane

The values of ET are known for several hundred solvents based on measurements of solvent-induced shifts with betaine dye used as the solvatochromic indicator. Based on such data, solvents can be divided into 3 groups: protic (E NT from 0.5 to 1), dipolar non-hydrogen donating (E NT from 0.3 to 0.5) and apolar (E NT from 0 to 0.3). The E NT values have a good linear correlation with light absorption, reaction rates, and chemical equilibria. In addition, the E NT values have a very good correlation with the Kosower’s polarity parameter, Z, for which there is also large amount of data available. Both sets of data can be converted using the following equation: E T = 0.752Z − 7.87

[3.1.2]

Gutman2,3 chose the reaction enthalpy of solvent with the reference acceptor (antimony pentachloride) to quantify Lewis-donor properties. The donor number, DN, is a dimensionless parameter obtained from negative values of reaction enthalpy. The data obtained from electrochemical and NMR studies were combined into one scale in which data are available for several hundred solvents. These data have a linear correlation with E NT according to the following equation:4

3.2 Methods of solvent manufacture

AN = −59.9 + 1850 . ET

69

[3.1.3]

The donor number is frequently used in various fields of polymer chemistry (see Chapter 10). Another classification based on acidity/basicity of solvents allows the division of solvents into six groups containing protic-neutral; protogenic; protophilic; aprotic-protophilic; aprotic-protophobic; and aprotic-inert.4 Snyder5,6 developed classification of solvents for chromatography which arranges solvents according to their chromatographic strength. It is classification based on the solvent’s ability to engage in hydrogen bonding or dipole interaction using the experimentally determined partition coefficients by Rohrschneider.7 Eight groups of solvents were defined based on cluster analysis. In addition to the usefulness of this classification in chromatography, it was found recently that it is also useful in the design of coatings which do not affect undercoated paints.8 Numerous other classifications and sets of data are available, such as those included in various databases on solvent toxicity, their environmental fate, combustion properties, explosive limits, etc.

REFERENCES 1 2 3 4 5 6 7 8

Ch Reichardt, Solvents and Solvents Effects in Organic Chemistry, VCH, Weinheim, 1988. V Gutman, G Resch, Lecture Notes on Solution Chemistry, Wold Scientific, Singapore, 1995. G Gritzner, J. Mol. Liq., 73,74, 487-500 (1997). Y Marcus, The Properties of Solvents, John Wiley & Sons, Chichester, 1999. L R Snyder, J. Chromatographic Sci., 16, 223 (1978). S C Rutan, P W Carr, J Cheong, J H Park, L R Snyder, J. Chromatography, 463, 21 (1989). L Rohrschneider, Anal. Chem., 45, 1241 (1973). I R Owen, US Patent 5,464,888, 3M, 1995.

3.2 OVERVIEW OF METHODS OF SOLVENT MANUFACTURE George Wypych ChemTec Laboratories, Inc., Toronto, Canada

Crude oil is the major raw material source for solvents. Aliphatic and aromatic hydrocarbons are produced by physical processes used in petrochemical industry. Other solvents are synthetic but their starting raw materials are usually products of the petrochemical industry. Figure 3.2.1 shows the main groups of materials produced by petrochemical industry from crude oil. Two observations are pertinent: the main goal of petrochemical industry is to convert crude oil to fuels. Solvents are only a small fraction of materials produced. Figure 3.2.2 shows that solvents are not only used directly as solvents but are also the building blocks in the manufacture of a large number of materials produced by organic chemistry plants. Desalting of crude oil is the first step in crude oil processing. It is designed to remove corrosive salts which may cause catalyst deactivation. After desalting, crude oil is subjected to atomospheric distillation. Figure 3.2.3 shows a schematic diagram of the crude oil distillation process. The raw material is heated to 400oC and separated into fractions on 30-50 fraction trays in distillation column. The diagram in Figure 3.2.3 shows the main fractions obtained from this distillation. The heavier fraction cannot be distilled under atmospheric

70

George Wypych

Figure 3.2.1. Schematic diagram of petroleum industry products and yields. [Reproduced from reference 1]

pressure therefore in the next step vacuum is applied to increase volatilization and separation. Certain fractions from the distillation of crude oil are further refined in thermal cracking (visbreaking), coking, catalytic cracking, catalytic hydrocracking, alkylation,


Figure 3.2.2. Organic chemicals and building block flow diagram. [Reproduced from reference 2]

Figure 3.3.3. Crude oil distillation. [Reproduced from reference 1]

71

72

George Wypych

Figure 3.3.4. Simplified catalytic cracking flow diagram. [Reproduced from reference 1]

isomerization, polymerization, catalytic reforming, solvent extraction, and other operations. Thermal cracking (visbreaking), uses heat and pressure to break large hydrocarbon molecules into lower molecular weight products. Most refineries do not use this process but use instead its replacement − catalytic cracking which gives a better yield of gasoline. Feedstock includes light and heavy oils from the crude oil distillation process. The cracking process occurs at a temperature of 550oC and under increased pressure. The cracking reaction is discontinued by mixing with cooler recycle stream. The mixture is stripped of lighter fractions which are then subjected to fractional distillation. Figure 3.2.4 is a schematic flow diagram of catalytic cracking. Hydrocracking is a somewhat different process which occurs under higher pressure and in the presence of hydrogen. It is used to convert fractions which are difficult to crack, such as middle distillates, cycle oils, residual fuel oils, and reduced crudes. Alkylation is used to produce compounds from olefins and isoparaffins in a catalyzed process. Isomerization converts paraffins to isoparaffins. Polymerization converts propene and butene to high octane gasoline. The application of these three processes has increased output and performance of gasoline. Catalytic reforming processes gasolines and naphthas from the distillation unit into aromatics. Four major reactions occur: dehydrogenation of naphthenes to aromatics, dehydrocyclization of paraffins to aromatics, isomerization, and hydrocracking.


73

In some cases mixtures of solvents are required to meet a particular requirement. For example, linear paraffins have a very low viscosity. Branched paraffins have high viscosity but very good low temperature properties and low odor. A combination of the two carried out in the conversion process (not by mixing) results in a solvent which has the desirable properties of both solvents: low viscosity and good low temperature properties.3 The recovery of pure aromatics from hydrocarbon mixtures is not possible using distillation process because the boiling points of many non-aromatics are very close to benzene, toluene, etc. Also, azeotropes are formed between aromatics and aliphatics. Three principle methods are used for separation: azeotropic distillation, liquid-liquid extraction, and extractive distillation. Three major commercial processes have been developed for separation: Udex, Sulpholane, and Arosolvan. Over 90% plants now use one of these processes. Each use an addition of solvent such as a mixture of glycols, tetramethylene sulfone, or N-methyl-2-pyrrolidone to aid in the extraction of aromatics. This occurs with high precision and efficiency. Pure benzene, toluene, and xylene are produced by these processes. These three are used for synthesis of several other important solvents. Benzene is used in the production of ethyl benzene (alkylation), cyclohexanone, cyclohexanol, cyclohexane, aniline (hydrogenation), acetone, nitrobenzene, and chlorobenzene. Toluene is used in the production of cresol and benzene. Xylene is the raw material for the production of ethyl benzene and the fractionation of the xylene mixture to isomers. Lower boiling fractions from the primary distillation are also used in the production of solvents. Ethylene is used to produce ethylene dichloride, ethylene glycol, ethanol, and ethyl benzene. Propylene is used to produce isopropyl alcohol. Halogenation, hydrohalogenation, alkylation, and hydrolysis reactions are used in these conversions. With different feedstock and methods of processing it is inevitable that there will be some differences between products coming from different feedstock sources and manufacturers. Over the years processes have been refined. Long practice, globalization of technology has occurred and restrictions have been imposed. Today, these differences are small but in some technological processes even these very small differences in solvent quality may require compensating actions. There are many other unitary operations which are used by organic chemistry plants to manufacture synthetic solvents. These include: alkoxylation (ethylene glycol), halogenation (1,1,1-trichloethane), catalytic cracking (hexane), pyrolysis (acetone and xylene), hydrodealkylation (xylene), nitration (nitrobenzene), hydrogenation (n-butanol, 1,6-hexanediol), oxidation (1,6-hexanediol), esterification (1,6-hexanediol), and many more. In the manufacture of oxygenated solvents, the typical chemical reactions are hydration, dehydration, hydrogenation, dehydrogenation, dimerization and esterification. For example methyl ethyl ketone is manufactured from 1-butene in a two step reaction. First, 1-butene is hydrated to 2-butanol then a dehydrogenation step converts it to methyl ethyl ketone. The production of methyl isobutyl ketone requires several steps. First acetone is dimerized producing diacetone alcohol which, after dehydration, gives mesityl oxide subjected in the next step to hydrogenation to result in the final product. Ethylene glycol is a product of the addition (ethylene oxide and ethanol) followed by esterification with acetic acid. 18% of all phenol production is converted to cyclohexanone.4 Some solvents are obtained by fermentation processes (e.g., ethanol, methanol).

74

George Wypych

Synthetic routes are usually quite complex. For example, the manufacture of N-methyl-2-pyrrolidone involves the reaction of acetylene with formaldehyde. The resulting but-2-ine-1,4-diol is hydrated to butane-1,4-diol and then dehydrated to γ-butyrolactone then reacted with monomethyl amine to give the final product. Strict process control is essential to obtain very high purity. New processes have been developed to produce solvents which are based on non-conventional materials (e.g., lactide and drying oil). The resultant solvent is non-volatile and useful in production of coatings, paints and inks.5 These new technological processes are driven by the need to reduce VOCs.

REFERENCES 1 2 3 4 5

EPA Office of Compliance Sector Notebook Project. Profile of the Petroleum Refining Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Profile of the Organic Chemical Industry. US Environmental Protection Agency, 1995. R J Wittenbrink, S E Silverberg, D F Ryan, US Patent 5,906,727, Exxon Research And Engineering Co., 1999. A M Thayer, Chem. Eng. News, 76, 16, 32-34 (1998). D Westerhoff, US Patent 5,506,294, 1996.

3.3 SOLVENT PROPERTIES George Wypych ChemTec Laboratories, Inc., Toronto, Canada

The purpose of this section is to provide analysis of properties of major groups of solvents. The data on individual solvents are included in a separate publication on CD-ROM as a searchable database containing 1145 the most common solvents. The information on properties of solvents is included in 110 fields containing chemical identification of solvent, physical chemical properties, health and safety data, and its environmental fate. Here, the analysis of this data is provided in a form of tables to show the range of properties for different groups of solvents and their strengths and weaknesses. For each group of solvents a separate table is given below. No additional discussion is provided since data are self-explanatory. The data are analyzed in the final table to highlight the best performance of various groups of solvents in different properties. By their nature, these data have a general meaning and for the exact information on particular solvent full data on CD-ROM should be consulted. For example, in the list of target organs, there are included all organs involved on exposure to solvents included in the group which does not necessary apply to a particular solvent. The data are given to highlight overall performance of entire group which may contain very diverse chemical materials. One very obvious method of use of this information is to review the list of carcinogens and mutagens to and relate them to actually used in particular application. These solvents should be restricted from use and possibly eliminated or equipment engineered to prevent exposure of workers and release to environment. The comparative tables also provide suggestions as to where to look for suitable substitutes based on physical chemical characteristics and potential health and environmental problems.

3.3 Solvent properties

75

Other application of this data is in selection of solvents for new products. The bulk of the data allows to analyze potential requirements critical for application and select group or groups which contain solvents having these properties. Further, based on their health and environmental characteristics suitable candidates can be selected. The bulk data are also very useful in constructing set of requirements for specification. Many specifications for industrial solvents are very simplistic, which is partially caused by the lack of data provided by manufacturers of these solvents. This may cause potential fear of future problems with such solvents since solvent replacement in formulated product is not always a trivial substitution. It is also possible that solvents for which inadequate data exist at the present moment will be more studied in future for their environmental and health impacts and will then require to be replaced. Also, incomplete specification means that solvents may contain undesirable admixtures and contaminations and were used on premisses that manufacturer considered its application for a particular purpose which was, in fact, never intended by solvent manufacturer for this particular application. These details should be obtained from manufacturers and relevant parameters included in the specification. It is known from experience of creation of the database of solvents that there are still large gaps in information which should be eliminated by future efforts. The practice of buying solvents based on their boiling point and specific gravity does not serve the purpose of selecting reliable range of raw materials. These and other aspects of created reference tables should be continuously updated in future to provide a reliable base of data which will be broadly used by industry. Application of full information allows to decrease quantity of solvents required for task and eliminate questionable materials and wastes due to solvent evaporation too rapid to make an impact on product properties at the time of its application. Also many sources of the problems with formulated products are due to various manifestations of incompatibility which can be eliminated (or predicted) based on solvent’s characteristics. 3.3.1 HYDROCARBONS 3.3.1.1 Aliphatic hydrocarbons Property

Value minimum

maximum

median

-11.7

285

124

-189

18

-75

-104

129

46

autoignition temperature, C

202

640

287

refractive index

1.29

1.46

1.41

specific gravity, g/cm

0.51

0.84

0.74

vapor density (air=1)

1

5.90

4.5

vapor pressure, kPa

0.00

1976

4.42

viscosity, mPa.s

0.21

1.58

0.46

15

40

21

o

boiling temperature, C o

freezing temperature, C o

flash point, C o

3

surface tension, mN/m

76

George Wypych

Value

Property

minimum

maximum

median

donor number, DN, kcal/mol

0

0

0

acceptor number, AN

0

1.6

0

30.9

31.1

31

8

10

9

specific heat, cal/K mol

41.24

93.41

60.33

heat of vaporization, cal/g

6.32

8.47

7.54

heat of combustion, MJ/kg

4.35

49.58

46.52

dielectric constant

1.8

2.15

2

Kauri-butanol number

22

56

32

21

165

81

6.8

8.2

7.4

Henry’s Law constant, atm/m mol

2.59E-4

4.56E-4

3.43E-4

evaporation rate (butyl acetate = 1)

0.006

17.5

0.45

threshold limiting value - 8h average, ppm

0.1

1000

300

maximum concentration (15 min exp), ppm

375

1000

500

LD50 oral, mg/kg

218

29820

2140

polarity parameter, ET(30), kcal/mol -4 o

coefficient of cubic expansion, 10 / C

o

aniline point, C Hildebrand solubility parameter, cal

1/2

-3/2

cm

3

route of entry

absorption, contact. ingestion, inhalation

target organs

blood, bone marrow, central nervous system, eye, gastrointestinal tract, heart, kidney, lymphatic system, liver, lung, nervous system, peripheral nervous system, respiratory system, skin, spleen, stomach, testes, thyroid

carcinogenicity

-

mutagenic properties theoretical oxygen demand, g/g

n-hexane 3.46

biodegradation probability

3.56

3.53

days-weeks

octanol/water partition coefficient

2.3

5.98

urban ozone formation

0.11

0.13

0.12

3.3.1.2 Aromatic hydrocarbons Property o

boiling temperature, C

Value minimum

maximum

median

74

288

168


77

Value

Property

minimum

maximum

median

-96

5.5

-31

-11

144

52


204

550

480

refractive index

1.43

1.61

1.5


0.71

1.02

0.87


2.8

4.8

4.1

vapor pressure, kPa

0.00

21.3

0.43

viscosity, mPa.s

0.58

6.3

1.04


24.3

36.8

28


0.1

10

5

acceptor number, AN

6.8

8.2

7.3

polarity parameter, ET(30), kcal/mol

32.9

34.8

34.7

8

10.7

8


32.44

63.48

52.41


8.09

10.38

10.13


41.03

43.5

41.49

dielectric constant

2.04

2.6

2.3

33

112

86

7

85

26

7.9

9.3

8.8


5.19E-3

3.8E-1

7.6E-3


0.006

5.1

0.16


0.3

100

50


6

150

125

LD50 oral, mg/kg

636

6989

4300

LD50 dermal, mg/kg

4400

17800

12400

o


flash point, C o

3

coefficient of cubic expansion, 10-4/oC

Kauri-butanol number o

aniline point, C Hildebrand solubility parameter, cal

1/2

-3/2

cm

3

route of entry


target organs

blood, bone marrow, central nervous system, eye, gastrointestinal tract, kidney, liver, lung, respiratory system, skin

carcinogenicity

benzene, styrene

mutagenic properties

benzene, ethylbenzene, xylene

78

George Wypych

Value

Property

minimum

maximum

median

biological oxygen demand, 5-day test, g/g

0.92

2.53

1.65

chemical oxygen demand, g/g

2.15

2.91

2.56

theoretical oxygen demand, g/g

2.41

3.29

3.17


days-to weeks, weeks


2.13

4.83


0.03

1.13

0.90

3.3.2 HALOGENATED HYDROCARBONS Value

Property

minimum

maximum

median

-40.6

253

87

-189

17

-36

-50

350

45


240

648

557

refractive index

1.20

1.63

1.43


0.9

3

1.6


1.8

33.4

6.5

vapor pressure, kPa

0.01

4000

10.2

viscosity, mPa.s

0.02

5.14

1.1


0.03

33.4

15.2

pH

4

8

7


0

4

3

acceptor number, AN

8.6

23.1

16.2


32.1

41.3

36.7


12.32

205

41.5


3.76

81.2

11.54


6.27

29.14

15.73

dielectric constant

1.0

8.93

2.32

31

500

90

5.9

10.75

8.3


3.6E-4

8.5E0

3E-2


0.9

14.5

1

o



flash point, C o

3

Kauri-butanol number Hildebrand solubility parameter, cal 3

1/2

cm

-3/2


79

Property

Value minimum

maximum

median


0.5

1000

100


20

1250

250

maximum concentration any time, ppm

5

500

400

LD50 oral, mg/kg

214

13000

1210

LD50 dermal, mg/kg

500

20000

8750

route of entry


target organs

central nervous system, eye, gastrointestinal tract, heart, kidney, liver, lung, respiratory system, skin

carcinogenicity

benzotrichloride, carbon tetrachloride,chloroform, 1,2-dibromomethane, 1,4-dichlorobenzene, 1,2-dichloroethane, Dowper, 1,1,2,2-tetrachlroethylene


benzotrichloride, carbon tetrachloride, chloroform, chloromethane, chlorodifluoromethane, dichloromethane, 1,2-dibromomethane, Freon MS-117 TE, Freon MS-178 TES, 1,1,2,2-tetrachloroethylene, 1,1,1-trichloroethane, 1,1,2-trichloroethylene, trifluoromethane


0


0.19

0.09

weeks


0.64

4.02

ozone depletion potential

0.00

1.1

0.8

global warming potential

0.25

11700

4600


0.00

0.09

0.01

3.3.3 NITROGEN-CONTAINING COMPOUNDS (NITRATES, NITRILES) Property o



flash point, C o

Value minimum

maximum

median

77

234

134

-112

6

-50

2

101

36


414

550

481

refractive index

1.34

1.55

1.39

specific gravity, g/cm3

0.78

1.21

0.87


1.4

4.2

3.1

80

George Wypych

Value

Property

minimum

maximum

median

vapor pressure, kPa

0.01

11.0

2.62

viscosity, mPa.s

0.34

1.96

0.77

dissociation constant, pKa

7.67

10.21

8.98


4.4

16.6

11.0

acceptor number, AN

14.8

20.5

17.7


41.5

46.7

43.6

22.46

30.38

27.25

9.5

12.3

10.5

1.15

2.3

2.1


1

100

25

LD50 oral, mg/kg

39

3800

510

heat of combustion, MJ/kg Hildebrand solubility parameter, cal

1/2

-3/2

cm


route of entry


target organs

blood, central nervous system, eye, kidney, liver, respiratory system, skin

carcinogenicity

acrylonitrile, 2-nitropropane


acrylonitrile, 2-nitropropane

biodegradation probability octanol/water partition coefficient

days-weeks -0.3

1.86

3.3.4 ORGANIC SULFUR COMPOUNDS Property o


Value minimum

maximum

median

37

287

142

freezing temperature, C

-116

32

-38

flash point, oC

-38

177

43

autoignition temperature, oC

206

470

395

refractive index

1.38

1.62

1.47

specific gravity, g/cm3

0.80

1.43

1.00


2.14

4.35

3.05

vapor pressure, kPa

0.00

19.00

1.05

viscosity, mPa.s

0.28

10.29

0.97


35.5

42.98

39.00


81

Value

Property

minimum

maximum

median

-1.54

15.3

13.6

2

41

29.8

acceptor number, AN

7.5

19.3

19.2


26.8

54.4

38.4


36.61

43

40.1

43.26

46.45

44.30

8.2

12.6

9.8


4.96E-8

4.85E-6

1.25E-6


0.005

0.026

0.013


0.1

10

0.5

LD50 oral, mg/kg

505

14500

1941

LD50 dermal, mg/kg

380

40000

20000

dissociation constant, pKa donor number, DN, kcal/mol

dielectric constant Hildebrand solubility parameter, cal

1/2

-3/2

cm

3

route of entry


target organs

central nervous system, eye, liver, lung

carcinogenicity

diethyl sulfate, dimethyl sulfate


diethyl sulfate, dimethyl sulfoxide, sulfolane


1.73


1.84

1.75

days-weeks


-1.35

2.28


0.07

0.23

0.15

3.3.5 MONOHYDRIC ALCOHOLS Property

Value minimum

maximum

median

64.55

259

155

-129

71

-38.6

11

156

67

231

470

295

1.277

1.539

1.42


0.79

1.51

0.81


1.10

5.50

3.0

vapor pressure, kPa

0.00

21.20

0.4

o



flash point, C o

autoignition temperature, C refractive index 3

82

George Wypych

Value

Property

minimum

maximum

median

viscosity, mPa.s

0.59

41.1

4.4


21.99

40.0

26.2

9.3

19.0

15.4

5

44

30

22.2

66.7

37.1

41

65.3

48.8

9

12.2

10.3


19.47

78.03

43.03


8.95

15.40

12.32


22.66

38.83

34.56

8.17

32.66

17.51

dissociation constant, pKa donor number, DN, kcal/mol acceptor number, AN polarity parameter, ET(30), kcal/mol -4 o


dielectric constant 1/2

-3/2

9.26

23

11.5

Henry’s Law constant, atm/m3 mol

4.1E-9

3.44E+1

3.1E-5


0.005

2.9

0.39


1

1000

100


4

500

125

LD50 oral, mg/kg

275

50000

2300

LD50 dermal, mg/kg

400

20000

3540

Hildebrand solubility parameter, cal

cm

route of entry


target organs

central nervous system, eye, kidney, liver, lung, lymphatic system, respiratory system, skin

carcinogenicity mutagenic properties

1-butanol, 2-butanol, 1-pentanol, 1-propanol

ethanol,

1-octanol,


0.41

2.37

1.5


1.5

2.97

2.46


1.5

2.9

2.59


days-weeks


-1.57

2.97


0.04

0.45

0.16


83

3.3.6 POLYHYDRIC ALCOHOLS Value

Property

minimum

maximum

median

171

327.3

214

-114

60

-4

85

274

152


224

490

371

refractive index

1.43

1.48

1.44


0.92

1.22

1.12


2.14

6.70

3.10

vapor pressure, kPa

0.00

0.32

0.01

21

114.6

54.65


33.1

48.49

44.13


14.1

15.1

14.5

19

20

19

acceptor number, AN

34.5

46.6

36.2


51.8

56.3

54.1


36.1

294

77.6


13.0

18.7

16.2


19.16

29.86

23.69

7.7

35.0

28.8

10.7

16.18

12.81


4.91E-13

2.3E-7

6E-8


0.001

0.01

0.01


1

25

10


25

50

LD50 oral, mg/kg

105

50000

16000

LD50 dermal, mg/kg

2000

225000

20000

o



flash point, C o

3

viscosity, mPa.s



1/2

-3/2

cm

3

route of entry


target organs

blood, eye, gastrointestinal tract, kidney, lymphatic system, liver, lung, respiratory system, skin, spleen

carcinogenicity

-

84

George Wypych

Property

Value minimum

maximum

tetraethylene glycol, trimethylene glycol


median

triethylene

glycol,


0.03

1.08

0.18


1.29

1.64

1.57


1.07

1.68

1.60


days-weeks


-0.92

-2.02


0.16

0.47

0.38

3.3.7 PHENOLS Property

Value minimum

maximum

median

182

245

202

-18

105

25

43

127

95


558

715

599

refractive index

1.52

1.60

1.54


0.93

1.34

1.02


3.2

4.4

3.7

vapor pressure, kPa

0.00

0.23

0.02

viscosity, mPa.s

3.5

11.55

9.4


9.1

10.85

10.3


11

o



flash point, C o

3

acceptor number, AN

44.8

50.4


50.3

60.8

53.3

Hildebrand solubility parameter, cal1/2 cm-3/2

8.7

12.1

10.6

3.84E-11

3.14E-9


5

5

5

LD50 oral, mg/kg

40

320000

810

LD50 dermal, mg/kg

950

1040

Henry’s Law constant, atm/m3 mol

route of entry



85

Value

Property

minimum

maximum

median

target organs

central nervous system, eye, respiratory system, skin

carcinogenicity

3-chlorophenol, o-chlorophenol


3-chlorophenol, o-chlorophenol


0.59

2.47

3.3.8 ALDEHYDES Value

Property

minimum

maximum

median

-21

253

162

-123

12.4

-86

-39

102

13


180

424

196

refractive index

1.33

1.62

1.44


0.70

1.25

0.85


1

4.5

2.5

vapor pressure, kPa

0.00

438

2.30

viscosity, mPa.s

0.32

5.4

1.32


23.14

41.1

32.00

o



flash point, C o

3


16

acceptor number, AN

12.8


5.53

10.33

6.77

8.33

11.7

10.85


0.1

100


0.3

150


0.2

0.3

LD50 oral, mg/kg

46

3078

100

LD50 dermal, mg/kg

270

16000

582

Hildebrand solubility parameter, cal

1/2

-3/2

cm


7.8

route of entry


target organs

eye, heart, liver, respiratory system, skin

carcinogenicity

acetaldehyde, formaldehyde, furfural

2

86

George Wypych

Value

Property

minimum


maximum

median

acrolein, formaldehyde, furfural


0.00


1.72


1.07


0.77

0.74

2.00

1.67

days-weeks


0.35

1.48


0.94

1.55

1.23

3.3.9 ETHERS Value

Property

minimum

maximum

median

34.4

289

104

-137

64

-58

-46

135

25


189

618

429

refractive index

1.35

1.57

1.42


0.71

1.21

0.89


1.5

6.4

4.0

vapor pressure, kPa

0.00

174.7

1.33

viscosity, mPa.s

0.24

1.1

0.42


17.4

38.8

24.8


-5.4

-2.08

-2.92

6

24

19

acceptor number, AN

3.3

10.8

8


16

43.1

36


28.77

60.22

45.86


5.99

13.1

10.57


34.69

38.07

36.58

2.2

13.0

4.5

7

10.5

9.2


54E-9

8.32E0

3.19E-4


0.004

11.88

8.14

o



flash point, C o

3


dielectric constant Hildebrand solubility parameter, cal 3

1/2

cm

-3/2


87

Property

Value minimum

maximum

median


1

1000

200


2

500

250

LD50 oral, mg/kg

72

30900

4570

LD50 dermal, mg/kg

250

2000

7600

route of entry


target organs

blood, central nervous system, eye, kidney, liver, respiratory system, skin

carcinogenicity

bis(chloromethyl) ether, chloromethyl methyl ether, 1,4-dioxane, epichlorohydrin, ethylene oxide, propylene oxide


diethyl ether, 1,4-dioxane, ethylene oxide, propylene oxide


0.06

0.48

0.19


1.74

1.75

1.75


1.07

2.95

2.21


days-weeks


-0.56

5.10


0.02

0.49

0.31

3.3.10 GLYCOL ETHERS Property

Value minimum

maximum

median

boiling temperature, oC

117

265

191

freezing temperature, oC

-148

14

-83

flash point, oC

27

143

85

autoignition temperature, oC

174

406

255

refractive index

1.39

1.53

1.43


0.83

1.11

0.95


3.00

8.01

5.25

vapor pressure, kPa

0.00

1.33

0.12

viscosity, mPa.s

0.7

20.34

3.3


25.6

42.0

28.5

acceptor number, AN

9

36.1

3

88

George Wypych

Value

Property

minimum

maximum

median

38.6

52

51

9.7

11.5

11.2


24.85

108

65.27


10.33

14.3

13.3


24.3

30.54

28.75

5.1

29.6

10.5

8.2

12.2

8.8


6.5E-10

7.3E-5

7.3E-8


0.001

1.05

0.37


5

100

25


150

LD50 oral, mg/kg

470




1/2

-3/2

cm

3

150 16500

6500

route of entry


target organs

blood, brain, central nervous system, eye, kidney, lymphatic system, liver, lung, respiratory system, skin, spleen, testes

carcinogenicity

diethylene glycol monobutyl ether, diethylene glycol dimethyl ether, 2-ethoxyethanol, ethylene glycol diethyl ether, ethylene glycol monophenyl ether, triethylene glycol dimethyl ether



0.12

1.18

0.71


1.69

2.20

1.85


1.07

3.03

2.17


days-weeks


-1.57

3.12


0.27

0.58

0.44

3.3.11 KETONES Property o


freezing temperature, C

Value minimum

maximum

median

56.1

306

147

-92

28

-55


89

Value

Property

minimum

maximum

median

-18

143

44


393

620

465

refractive index

1.35

1.55

1.41


0.74

1.19

0.82


2

4.9

3.5

vapor pressure, kPa

0.00

30.8

1.1

viscosity, mPa.s

0.30

2.63

0.68


22.68

35.05

25.50


-8.3

24.2

20.5


11

18

17

o

flash point, C o

3

acceptor number, AN

12.5


36.3

42.2

39.8

coefficient of cubic expansion, 10-4/oC

9.7

13

13


29.85

58.22

51.0


7.48

12.17

9.94


26.82

40.11

36.35

11.98

20.56

16.1

7.54

11.0

9.2


4.4E-8

2.7E-4

8.7E-5


0.02

6.6

0.83


5

750

50


75

1000

300


5


1/2

-3/2

cm

3

LD50 oral, mg/kg

148

5800

2590

LD50 dermal, mg/kg

200

20000

6500

route of entry


target organs

central nervous system, eye, kidney, liver, lung, peripheral nervous system, respiratory system, skin, stomach, testes

carcinogenicity mutagenic properties biological oxygen demand, 5-day test, g/g

diacetone alcohol, methyl isopropyl ketone 0.68

2.03

1.37

90

George Wypych

Value

Property

minimum

maximum

median


1.92

2.88

2.31


1.67

2.93

2.44


days-weeks


-1.34

2.65


0.01

0.65

0.15

3.3.11 ACIDS Value

Property

minimum

maximum

median

20

337

164

-83

137

-3

37

140

100

298

539

380

1.285

1.551

1.421


0.9

1.83

1.08


0.7

5.0

3.3

vapor pressure, kPa

0.00

410

0.08

viscosity, mPa.s

0.25

23.55

2.82


27.4

37.6

33


0.23

4.88

4.25


2.3

20

10.5

acceptor number, AN

18.5

105

52.9


43.9

57.7

54.4


2367

29.42

2.612

4.8

5.58

4.80

6.17

58.5

40.5

9.79

15.84

12.29


1.26E-8

4.4E-5

1.67E-7


0.00

1.34

0.3


1

10

4


10

15

10

LD50 oral, mg/kg

200

74000

3310

o



flash point, C o


heat of vaporization, cal/g dielectric constant Hildebrand solubility parameter, cal

1/2

-3/2

cm

3


91

Property

Value minimum

maximum

median

route of entry


target organs

eye, kidney, liver, respiratory system, skin

carcinogenicity

-


formic acid


0.2

0.65


0.36

1.09


0.35

1.07

biodegradation probability octanol/water partition coefficient

0.67

days-weeks -0.17


+1.88 0-0.09

3.3.12 AMINES Property

Value minimum

maximum

median

-33

372

152

-115

142

-6

-37

198

55


210

685

410

refractive index

1.32

1.62

1.48


0.7

1.66

1.02


0.54

10.09

3.2

vapor pressure, kPa

0.00

1.013

0.13

viscosity, mPa.s

0.13

4000

3.15


19.11

48.89

32.43


8.96

11.07

10.78

pH

7.2

12.1

11


24

61

33.1

acceptor number, AN

1.4

39.8

18.8


32.1

55.8

42.2


30.4

74.1

53.4


5.65

16.13

8.26

o



flash point, C o

3


30.22

92

George Wypych

Value

Property

minimum

maximum

median

2.42

37.78

29.36

7.4

15.5

10.5


1.7E-23

3.38E+1

1.56E-8


0.001

3.59

0.06


0.1

100

5


6

35

15


5


1/2

-3/2

cm

3

LD50 oral, mg/kg

100

12760

470

LD50 dermal, mg/kg

64

8000

660

route of entry


target organs

eye, kidney, lymphatic system, liver, lung, respiratory system, skin, testes

carcinogenicity

acetamide, p-chloroaniline, N,N-dimethylformamide, hydrazine, N-nitrosodimethylamine, o-toluidyne


dimethylamine, ethylene diamine tetracetic acid, methylamine, N-methylpyrrolidone, N-nitrosomethyl amine, tetraethylene pentamine


0.01

2.24

0.84


1.28

1.9

1.53


0.65

2.85

1.8


days-weeks


-1.66

1.92


0.00

0.51

0.21

3.3.13 ESTERS Property

Value minimum

maximum

median

32

343

165

-148

27.5

-54

-19

240

64


252

505

400

refractive index

1.34

1.56

1.44

0.81

1.38

0.92

o



flash point, C o

3



93

Value

Property

minimum

maximum

median


2.5

9.60

5.2

vapor pressure, kPa

0.00

64.0

0.27

viscosity, mPa.s

0.42

32.7

1.07


23.75

41.39

28.6


10.68

13.3

12.0

pH

5

7

7


11

23.7

16.3

acceptor number, AN

6.7

18.3

16.3


36.7

48.6

40.9

-4 o


8.76

10.3


31.54

119

46.9


7.72

21.8

10.04


18.5

36.35

28.19

dielectric constant

4.75

64.9

64.0

62

1000

1000

7.34

12.6

8.8


9.9E-8

1.9E-2

3.6E-4


0.001

11.8

0.22


0.2

400

100


2

310

150

LD50 oral, mg/kg

500

42000

5600

LD50 dermal, mg/kg

500

20000

5000

Kauri-butanol number Hildebrand solubility parameter, cal1/2 cm-3/2 3

route of entry


target organs

blood, brain, central nervous system, eye, gastrointestinal tract, lung, respiratory system, skin, spleen

carcinogenicity

ethyl acrylate, vinyl acetate


methyl ester of butyric acid, γ-butyrlactone, dibutyl phthalate, 2-ethoxyethyl acetate, ethyl acetate, ethyl propionate, ethylene glycol methyl ether acetate, methyl propionate, n-propyl acetate


0.25

1.26

0.6


1.11

2.32

1.67


1.09

2.44

1.67

94

George Wypych

Value

Property

minimum


maximum

median

days-weeks


-0.56

+3.97


0.02

0.42

0.08

3.3.14 COMPARATIVE ANALYSIS OF ALL SOLVENTS Value

Property

minimum

maximum

range

CFCs

PHA

-40.6-372

CFCs

amines

-189-142

aliphatic HC

CFCs (none)

-104-350

glycol ethers

phenols

174-715

CFCs

halogenated

1.20-1.63


aliphatic HC

CFCs

0.51-3


aldehydes

CFCs

1-33.4

vapor pressure, kPa

many

CFCs

0.00-4000

viscosity, mPa.s

CFCs

PHA

0.02-114.6


CFCs

PHA

0.03-48.49


ethers

alcohols

-8.3-19.00

pH

acids

amines

1-14


hydrocarbons

amines

0-61

acceptor number, AN

hydrocarbons

acids

0-105

ethers

alcohols

16-65.3

alcohols

ethers

7-14.5


CFCs

PHA

12.32-294


CFCs

halogenated

3.76-81.2


CFCs

aliphatic HC

6.57-44.58

dielectric constant

CFCs

esters

1.0-64.9

aliphatic HC

esters

22-1000

aromatic HC

aliphatic HC

7-165

CFCs

alcohols

5.9-23


amines

alcohols

1.7E-23-34.4


many

aliphatic HC

0-17.5

o



flash point, C o




Kauri-butanol number o

aniline point, C Hildebrand solubility parameter, cal 3

1/2

cm

-3/2


Property

95

Value minimum

maximum

range


several

several

0.1-1000


aldehydes

CFCs

0.3-1250

LD50 oral, mg/kg

aldehydes

phenols

46-320000

amines

alcohols

64-225000

LD50 dermal, mg/kg route of entry


target organs

blood, brain, bone marrow, central nervous system, eye, gastrointestinal tract, heart, kidney, lymphatic system, liver, lung, peripheral nervous system, respiratory system, skin, spleen, stomach, testes, thyroid

carcinogenicity

some in the following groups: aromatic hydrocarbons, halogenated hydrocarbons, nitrogen-containing compounds, organic sulfur compounds, phenols, aldehydes, ethers, amines, esters


each group contains some species


CFCs

aliphatic HC

0-3.56


days-weeks in the most cases

ozone depletion potential

CFCs

global warming potential

CFCs


CFCs

aldehydes

0-1.55

HC - hydrocarbons, PHA - polyhydric alcohols

The comparative chart allocates for each group the highest and the lowest position in relationship to their respective values of particular parameters. The chart allows to show that the fact of having many good solvent properties does not warrant that solvent is suitable for use. For example, CFCs have many characteristics of good solvents but they are still eliminated from use because they cause ozone depletion and are considered to be a reason for global warming. On the other hand, esters do not appear on this chart frequently but they are very common solvents. The chart also shows that solvents offer very broad choice of properties, which can be selected to satisfy any practical application.

96

Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg

3.4 TERPENES Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg Institut fòr Allgemeine Hygiene und Umwelthygiene University of Tòbingen, Tòbingen, Germany

3.4.1 DEFINITIONS AND NOMENCLATURE Terpenes are natural products. Terpenoids enlarge this division to include natural or synthetic derivatives. The structures of terpenes or terpenoids varies widely and are classified according to various aspects, e.g. number of isoprene units (C10 monoterpenes, C15 sesquiterpenes, C20 diterpenes, C25 sesterterpenes, C30 triterpenes, C40 tetraterpenes) or division in acyclic, mono-, bi-, tri-, tetra-, pentacyclic terpenes.1,2 Often terpenes are named by their trivial names, e.g. d-limonene. 3.4.2 OCCURRENCE The occurrence of terpenes is ubiquitous. Natural terpenes are found in plants and animals in minute amounts. Especially in higher plants, terpenes characterize the type of plant (chemotaxonomy): mono- and sesquiterpenes in essential oils, sesqui-, di-, triterpenes in balsams and resins, tetraterpenes in pigments and polyterpenes in latexes.1,3,4,5 Therefore, terpenes are often emitted from natural products such as citrus fruits or trees, e.g. conifers. Terpenes are components of various products: e.g. tobacco smoke, wax pastes (furniture and floor polishes etc.), liquid waxes (floor polishes etc.), cleansers (detergents etc.), polishes, dyes and varnishes, synthetic resins, so-called “natural” building products, deodorants, perfumes, softeners, air fresheners, foods, beverages, pharmaceutical products (e.g. camomile oil, eucalyptus oil).1,3,4,5 In these products terpene compounds such as geraniol, myrcene (beta-myrcene), ocimene, menthol, alpha-pinene, beta-pinene, d-limonene, 3-carene, cineole, camphene or caryophyllene can be detected.1,3,4,5 Often terpenes may be included as additives, e.g. food additives licensed by the FDA. Terpenes detected in indoor air are mainly the monoterpenes alpha-, beta- pinenes, 3-carene and d-limonene which occur primarily in conifer products.5 Some of the monoterpenes may be converted into well-known epoxides and peroxides with high allergic potential.5 Several products containing terpenes are more highly refined which influences quality and quantities of terpenes in these products: e.g. in oil of turpentine or in resin components. The quantity of monoterpenes is essentially influenced by the composition of the raw materials, e.g. d-limonene dominates in agrumen oils as citrus oil products.5 Terpene products are often associated with “natural positive” properties, e.g. attributes such as “biological, positive health effects and good biodegradability” which are often neither substantiated nor proven. Indoor concentrations of some terpenes, e.g., d-limonene and pinene, are highest in the group of VOCs.6 3.4.3 GENERAL Terpenes are synthesized in chloroplasts, mitochondria and microsomes of plants or in the liver of animals. Typical biosynthesis pathways of terpenes are well-known, e.g. via

3.4 Terpenes

97

decarboxylation, isomerization and acetyl-CoA-processes.1 Degradation of terpenes is possible by microorganisms, e.g. Pseudomonas and Aspergillus sp., in plants and in animals.1 In terpene products, terpenes exist as two enantiomers in different mixture ratios. Enantiomers are associated with characteristic odour (e.g. d-limonene in orange-oil).5 Odors of terpenes are essential criteria in the classification of terpenes. Some terpenes can be smelt in extremely low concentrations.5 Threshold limit values are in the range of µg/m3 in indoor air.5 This applies primarily to monoterpenes pinenes, d-limonene, carenes and sesquiterpenes longifolenes and caryophyllenes. 3.4.4 TOXICOLOGY Most terpenes show low acute oral toxicity and low dermal toxicity. Contact dermatitis is the most common symptom described as a result of exposure to terpenes. Other allergic reactions occur more rarely: e.g. allergic rhinitis or allergic bronchial asthma. The most common products with an allergic potential (contact dermatitis) are oils of turpentine.5 Older turpentine products show higher allergic potential than freshly distilled products. Turpentines have now been replaced by other less toxic petrochemical products. Many terpenes or metabolites are well-known contact allergens causing allergic dermatitis, e.g. d-limonene or oxygenated monocyclic terpenes which are produced by autoxidation of d-limonene.7 Normally the highest allergic potential is associated with photo-oxidants (e.g. peroxides, epoxides) which are formed from terpenes. The symptoms of allergic dermatitis disappear if dermal contact to the causative terpene allergens is removed. Exposure to the monoterpenes (alpha-pinene, beta-pinene and 3-carene) showed no major changes in lung function, but showed chronic reaction in the airways (reduced lung function values which persist between shifts) in workers of joinery shops.8 In studies of dwellings, bronchial hyper-responsiveness could be related to indoor air concentrations of d-limonene.9 Other studies did not find significant changes in the respiratory tract.10,11 Nevertheless, these studies postulate effects of metabolites of terpenes (e.g. pinenes) as relevant causative agents. It is suggested in some cases that “Multiple Chemical Sensitivity” (MCS) may be attributed to increased values of terpenes and aromatic hydrocarbons.12 Often mixtures of terpene products or so-called “natural products” show allergic effects, e.g. fragrant mixtures containing d-limonene,7 tea tree oil,13 oilseed rape.14 Consumer products such as deodorants or perfumes also contain terpenes with allergic potential.15,16 3.4.5 THRESHOLD LIMIT VALUES Relevant threshold limit values for terpenes are rare because of a lack of basic information about specific terpene products and by-products on the one hand, and occupational and environmental exposures on the other hand.5 The threshold limit values which have been documented the best concern oil of turpentine. A MAK-value of 100 ppm is defined in German regulations and noted to be dermally sensitive.17 For other terpenes, such as d-limonene which is also classified as dermally sensitive, it has not yet been possible to establish a MAK-value because of a lack of information of their effects on animals or humans. With terpenes, as is often the case, aggregate concentration parameters are used as limit values such as the minimum level goals recommended by the former German Federal Health Authority.18 These suggested minimum values bear in mind actual levels detected in indoor areas.

98

Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg

Although Mohr5 quotes values of 30 µg/m3 for terpene aggregate and 15 µg/m3 for a single terpene compound from a scheme of Seifert, he considers values of 60 µg/m3 for terpene aggregate to be more appropriate according to his experience. He also discusses aggregate values of 200 µg/m3 (these exceed the olfactory threshold level) as posing possible health hazards in individual cases. These concentration limits only concern monoterpenes (pinenes, d-limonene, 3-carene) not other terpene products (e.g. sesquiterpenes).5 Table 3.4.1 Selected examples of terpenes Substance group

Examples

Common occurrence

Selected properties

geraniol

in essential oils, perfume products and luxury foods, production from other terpene products, e.g. beta-pinene

acyclic, double unsaturated alcohol, several possible reactions, occurrence as esters, typical rose odor

myrcene

naturally occurring in plant oils and organisms, industrial production

very reactive, pleasant odor, part of many synthetic reactions (synthesis of other terpenes)

ocimene

in essential oils and perfume products

pleasant odor, sensitive to oxidation

p-menthane

in essential oils, e.g. eucalyptus fruits

typical odor (fennel)

p-cimene

perfume and soap products (musk perfumes), in various production processes, e.g., sulfite leaching of wood

typical odor (aromatic hydrocarbons), inflammable

Bicyclic monoterpenes

pinenes

in essential oils and conifer products, industrial production and use, e.g. in fragrance and flavor industry

typical odor (turpentine), softening agent

Acyclic sesquiterpenes

farnesol

in essential oils, in perfume and soap products

typical odor, sensitive to oxidation, heat and light

Monocyclic sesquiterpenes

bisabolenes

in various oils, e.g. myrrh and limete oil, in perfume and fragrance products

balsamic odor

Bicyclic sesquiterpenes

caryophyllenes

in essential oils, as perfume and fragrance products, chewing gum, synthesis of other perfumes

typical odor (clove)

Tricyclic sesquiterpenes

longifolene

in essential oils, e.g. turpentine, solvent additive, production of perfumes

colorless, oily liquid

Acyclic monoterpenes

Monocyclic monoterpenes

3.4 Terpenes

99


6 7 8

9 10

11 12 13 14

15

16

17 18

Ullmann´s Encyclopedia of Industrial Chemistry, 1998. O.W. Thiele, Lipide, Isoprenoide mit Steroiden, Thieme Verlag, Stuttgart, 1979. Römpp Lexikon, Encyclopedia of Chemistry. Chemie-Lexikon, Thieme, 1998. Seifert, B., Esdorn, H., Fischer, M., Rüden, H., Wegner, J. (eds.), Indoor air ´87. Proceeding of the 4th International Conference on Indoor Air Quality and Climate. Institute for Soil, Water and Air Hygiene, 1987. Mohr, S., Neue Lasten durch ökologische Baustoffe ? Vorkommen und Bewertung von Terpenen in der Innenraumluft. In Behrends, B. (ed.), Gesundes Wohnen durch ökologisches Bauen, Hannoversche Ärzte-Verlags-Union, 61-101, 1998. Lance, A., Wallace, Ph.D., Volatile organic compounds, In Samet, J.M., and Spengler, J.D. (eds.): Indoor air pollution: a health perspective. The Johns Hopkins University Press, 1991. Karlberg, A.T., DoomsGoossens, A., Contact allergy to oxidized d-limonene among dermatitis patients. Contact Dermatitis, 36, 201-206, 1997. Eriksson, K.A., Levin, J.O., Sandstrom, T., Lindstrom-Espeling, K., Linden, G., Stjernberg, N.L., Terpene exposure and respiratory effects among workers in Swedish joinery shops, Scan. J. Work. Environ. Health, 23, 114-120, 1997. Norback, D., Bjornsson, E., Janson, C., Widstrom, J., Boman, G., Asthmatic symptoms and volatile organic compounds, formaldehyde, and carbon dioxide in dwellings, Occup. Environ. Med., 52, 388-395, 1995. Falk, A.A., Hagberg, M.T., Lölf, A.E. , Wigaeus-Hjelm, E.M., Zhiping, W., Uptake, distribution and elimination of pinene in man after exposure by inhalation, Scand. J. Work Environ. Health, 16, 372-378, 1990. Eriksson, K., Levin, J.O., Identification of cis- and trans-verbenol in human urine after occupational exposure to terpenes. Int. Arch. Occup. Environ. Health, 62, 379-383, 1990. Ring, J., Eberlein-Konig, B., Behrendt, H., “Eco-syndrome”- “Multiple Chemical Sensitivity” (MCS), Zbl. Hyg. Umweltmed., 202, 207-218, 1999. Rubel, D.M., Freeman, S., Southwel, I.A., Tea tree oil allergy: what is the offending agent ? Report of three cases of tea tree oil allergy and review of the literature. Austral. J. Dermatol., 39, 244-247, 1998. McEwan, M., MacFarlane-Smith, W.H., Identification of volatile organic compounds emitted in the field by oilseed rape (Brassica napus ssp. Oleifera) over the growing season. Clin. Exp. Allergy, 28, 332-338, 1998. Rastogi, S.C., Johanson, J.D., Frosch, P., Menne, T., Bruze, M., Lepoittevin, J.P., Dreier, B., Andersen, K.E., White, I.R., Deodorants on the European market: quantitative chemical analysis of 21 fragrances, Contact Dermatitis, 38, 29-35, 1998. Johanson, J.D., Rastogi, S.C., Andersen, K.E., Menne, T., Content and reactivity to product perfumes in fragrance mix positive and negative eczema patients. A study of perfumes used in toileries and skin-care products, Contact Dermatitis, 36, 291-296, 1997. DFG (Deutsche Forschungsgemeinschaft), MAK- und BAT-Werte-Liste 1999, DFG. WILEY-VCH, 1999. Anonymous, Bewertung der Luftqualität in Innenräumen. Evaluation of Indoor Air Quality, Bundesgesundhbl., 3/93, 117-118, 1993.

4

General Principles Governing Dissolution of Materials in Solvents 4.1 SIMPLE SOLVENT CHARACTERISTICS Valery Yu. Senichev, Vasiliy V. Tereshatov Institute of Technical Chemistry Ural Branch of Russian Academy of Sciences, Perm, Russia

Polymer dissolution is a necessary step in many of the polymer processing methods, such as blending, separation, coating, casting, etc. The developments in physical chemistry of non-electrolyte solutions relate the capabilities of solvents to dissolve materials with their physical properties. The relationships were developed within the framework of the concept of solubility parameters. 4.1.1 SOLVENT POWER The usual problem of polymer engineering is a selection of proper solvent(s) for a given polymer. This selection implies that the solvent must form with polymer a thermodynamically stable mixture in the whole range of concentrations and temperatures. Such choice is facilitated by use of numerical criterion of a solvent power. Solvent power might be taken from the thermodynamic treatment (for example, a change of Gibbs’ free energy or chemical potentials of mixing of polymer with solvent) but these criteria depend not only on the solvent properties but also on polymer structure and its concentration. For this reason, various approaches were proposed to estimate solvent power. Kauri-butanol value, KB is used for evaluation of dissolving ability of hydrocarbon solvents. It is obtained by titration of a standard Kauri resin solution (20 wt% in 1-butanol) with the solvent until a cloud point is reached (for example, when it becomes impossible to read a text through the solution). The amount of the solvent used for titration is taken as KB value. The relationship between KB and solubility parameter, δ, fits the following empirical dependence:1 δ = 12.9 + 0.06KB

[4.1.1]

102

Valery Yu. Senichev, Vasiliy V. Tereshatov

The KB value is primarily a measure of the aromaticity of solvents. Using KB value, it is possible to arrange solvents in sequence: aliphatic hydrocarbons < naphthenic hydrocarbons < aromatic hydrocarbons. Dilution ratio, DR, is used to express the tolerance of solvents to diluents, most frequently, toluene. DR is the volume of a solvent added to a given solution that causes precipitation of the dissolved resin. This ratio can characterize the compatibility of a diluent with a resin solution in primary solvent. When compatibility is high, more diluent can be added. Only a multi-parameter approach provides a satisfactory correlation with solubility parameters.2-3 DR depends on the polymer concentration. With polymer concentration increasing, DR increases as well. Temperature influences DR in a similar manner. Determination of DR must be performed at standard conditions. DR can be related to the solubility parameters but such correlation depends on concentration. Aniline point, AP, is the temperature of a phase separation of aniline in a given solvent (the volume ratio of aniline : solvent = 1:1) in the temperature decreasing mode. AP is a critical temperature of the aniline - solvent system. AP can be related to KB value using the following equations: At KB50 5   .  358 − Tb  KB = 177.7 − 10.6ρ − 0.249AP + 010 9  

[4.1.3]

where: Tb

a solvent boiling point.

AP depends on the number of carbon atoms in the hydrocarbon molecule. AP is useful for describing complex aromatic solvents. The solvent power can also be presented as a sum of factors that promote solubility or decrease it:4 S =H + B − A −C −D

[4.1.4]

where: H B A C D

a factor characterizing the presence of active sites of opposite nature in solvent and polymer that can lead to formation of hydrogen bond between polymer and solvent a factor related to the difference in sizes of solute and solvent molecules a factor characterizing solute “melting” a factor of the self-association between solvent molecules a factor characterizing the change of nonspecific cohesion forces in the course of transfer of the polymer molecule into solution.

The equations for calculation of the above-listed factors are as follows: B = (α + b )

Vm (1 − ϕ p ) Vs

[4.1.5]

4.1 Simple solvent characteristics

D=

2 2 Vm δ′p − δ′s ) (1 − ϕ p ) ( RT

 K (1 − ϕ p )  H = ln 1 +  Vs   C = ln

103

[4.1.6]

[4.1.7]

1 + (K pp / Vm )

1 + (K pp ϕ p / Vm )

[4.1.8]

where: α

constant depending on the choice of the equation for the entropy of mixing. Usually it equals 0.5. constant depending on the structure of solvent. b=1 for unstructured solvents, b=-1 for solvents with single H-bond (e.g. alcohols) and b=-2 for solvents with double H-bonds chains such as water. the molar volume of a repeating segment the molar volume of solvent the volume fraction of polymer modified solubility parameters without regard to H-bonds. the stability constant of the corresponding solvent-polymer hydrogen bond the constant of the self-association of polymer segments

b

Vm Vs ϕp δp, δs K Kpp

Several polymers such as polyethylmethacrylate, polyisobutylmethacrylate and polymethylmethacrylate were studied according to Huyskens-Haulait-Pirson approach. The main advantage of this approach is that it accounts for entropy factors and other essential parameters affecting solubility. The disadvantages are more numerous, such as lack of physical meaning of some parameters, great number of variables, and insufficient coordination between factors influencing solubility that have reduced this approach to an approximate empirical scheme. 4.1.2 ONE-DIMENSIONAL SOLUBILITY PARAMETER APPROACH The thermodynamic affinity between components of a solution is important for quantitative estimation of mutual solubility. The concept of solubility parameters is based on enthalpy of the interaction between solvent and polymer. Solubility parameter is the square root of the cohesive energy density, CED: δ = (CED )

1/ 2

 ∆E i =   Vi

1/ 2

  

[4.1.9]

where: ∆Ei Vi

cohesive energy molar volume

Solubility parameters are measured in (MJ/m3)1/2 or (cal/sm3)1/2 (1 (MJ/m3)1/2=2.054 (cal/sm3)1/2). The cohesive energy is equal in magnitude and opposite in sign to the potential energy of a volume unit of a liquid. The molar cohesive energy is the energy associated with all molecular interactions in one mole of the material, i.e., it is the energy of a liquid relative to its ideal vapor at the same temperature (see Chapter 5).

104


δis a parameter of intermolecular interaction of an individual liquid. The aim of many studies was to find relationship between energy of mixing of liquids and their δ. The first attempt was made by Hildebrand and Scatchard5,6 who proposed the following equation: ∆U

m

 ∆E 1/ 2  ∆E 1 2  −  = (x 1V1 + x 2V2 )  V  V1   2

1/ 2

  

2

  ϕ1 ϕ 2 = 

= (x 1V1 + x 2V2 )(δ1 − δ 2 ) ϕ1 ϕ 2 2

[4.1.10]

where: ∆Um x1, x2 V1, V2 ϕ 1, ϕ 2

internal energy of mixing, that is a residual between energies of a solution and components, molar fractions of components molar volumes of components volume fractions of components

The Hildebrand-Scatchard equation became the basis of the Hildebrand theory of regular solutions.5 They interpreted a regular solution as a solution formed due to the ideal entropy of mixing and the change of an internal energy. The assumed lack of the volume change makes an enthalpy or heat of mixing equated with the right members of the equation. The equation permits calculation heat of mixing of two liquids. It is evident from equation that these heats can only be positive. Because of the equality of of components, ∆Hm=0. The free energy of mixing of solution can be calculated from the equation ∆G m

 ∆E 1/ 2  ∆E 1 2  −  = (x 1V1 + x 2V2 )   V1   V2

−T∆S id = V(δ1 − δ 2 ) ϕ1 ϕ 2 − T∆S 2

1/ 2

  

2

  ϕ1 ϕ 2 −  [4.1.11]

The change of entropy, ∆Sid, is calculated from the Gibbs equation for mixing of ideal gases. The calculated values are always positive. ∆S id = −R(x 1 ln x 1 + x 2 ln x 2 )

[4.1.12]

where: R

gas constant

Considering the signs of the parameters ∆Sid and ∆Hm in Eq. [4.1.10], the ideal entropy of mixing promotes a negative value of ∆Gm, i.e., the dissolution and the value of ∆Hm reduces the ∆Gm value. It is pertinent that the most negative ∆Gm value is when ∆Hm=0, i.e., when δ of components are equal. With these general principles in mind, the components with solubility parameters close to each other have the best mutual solubility. The theory of regular solutions has essential assumptions and restrictions.7 The Eq. [4.1.10] is deduced under assumption of the central role of dispersion forces of interaction between components of solution that is correct only for the dispersion forces. Only in this case it is possible to accept that the energy of contacts between heterogeneous molecules is a geometric mean value of energy of contacts between homogeneous molecules:


105

ε*12 = ε*11 ε*22

[4.1.13]

where: ε ∗ιι

potential energy of a pair of molecules

This assumption is not justified in the presence of the dipole-dipole interaction and other more specific interactions. Therefore the theory of regular solutions poorly suits description of the behavior of solutions of polar substances. Inherent in this analysis is the assumption of molecular separation related to molecular diameters which neglects polar or specific interactions. The theory also neglects volume changes on dissolution. This leads to a disparity (sometimes very large) between internal energy of mixing used in the theory and the constant pressure enthalpy measured experimentally. The correlation between these values is given by equation:

( ∆H ) = ( ∆U ) m

where:

(

+ T (∂p / ∂T )V ∆V m

m

p

V

)

[4.1.14]

p

( ∂p / ∂T )V thermal factor of pressure which has value of the order 10-14 atm/degree for solutions and liquids.

Therefore, even at small changes of volume, the second term remains very large and brings substantial contribution to the value of (∆Hm)p. For example, for a system benzene (0.5 mol) - cyclohexane (0.5 mol): ∆V m = 0.65 cm 3 ,

( ∆H ) m

p

= 182 cal,

( ∆U ) m

V

= 131 cal

The theory also assumes that the ideal entropy is possible for systems when ∆Hm≠0. But the change of energy of interactions occurs in the course of dissolution that determines the inevitable change of entropy of molecules. It is assumed that the interactive forces are additive and that the interactions between a pair of molecules are not influenced by the presence of other molecules. Certainly, such an assumption is simplistic, but at the same time it has allowed us to estimate solubility parameters using group contributions or molar attractive constants (see Subchapter 5.3). The solubility parameter δ is relative to the cohesion energy and it is an effective characteristic of intermolecular interactions. It varies from a magnitude of 12 (MJ/m3)1/2 for nonpolar substances up to 23 (MJ/m3)1/2 for water. Knowing δ of solvent and solute, we can estimate solvents in which particular polymer cannot be dissolved. For example, polyisobutylene for which δis in the range from 14 to 16 (MJ/m3)1/2 will not be dissolved in solvents with δ=20-24 (MJ/m3)1/2. The polar polymer with δ=18 (MJ/m3)1/2 will not dissolve in solvents with δ=14 or δ=26 (MJ/m3)1/2. These are important data because they help to narrow down a group of solvents potentially suitable for a given polymer. However, the opposite evaluation is not always valid because polymers and solvents with the identical solubility parameters are not always compatible. This limitation comes from integral character of the solubility parameter. The solubility depends on the presence of functional groups in molecules of solution components which are capable to interact with each other and this model does not address such interactions. The latter statement has become a premise for the development of the multi-dimensional approaches to solubility that will be the

106


subject of the following subchapter. The values of solubility parameters are included in Table 4.1.1. Table 4.1.1 Solubility parameters and their components (according Hansen’s approach) for different solvents

Solvent

V1, Kmol/m3

δ, (MJ/m3)1/2

δd, (MJ/m3)1/2

δp, (MJ/m3)1/2

δh, (MJ/m3)1/2

Alkanes n-Butane

101.4

14.1

14.1

0

0

n-Pentane

116.2

14.3

14.3

0

0

n-Hexane

131.6

14.8

14.8

0

0

n-Heptane

147.4

15.1

15.1

0

0

n-Octane

163.5

14.0

14.0

0

0

n-Nonane

178.3

15.4

15.4

0

0

n-Decane

195.9

15.8

15.8

0

0

n-Dodecane

228.5

16.0

16.0

0

0

Cyclohexane

108.7

16.7

16.7

0

0

Methylcyclohexane

128.3

16.0

16.0

0

0.5

Aromatic hydrocarbons Benzene

89.4

18.7

18.4

1.0

2.9

Toluene

106.8

18.2

18.0

1.4

2.0

Naphthalene

111.5

20.3

19.2

2.0

5.9

Styrene

115.6

19.0

17.8

1.0

3.1

o-Xylene

121.2

18.4

17.6

1.0

3.1

Ethylbenzene

123.1

18.0

17.8

0.6

1.4

Mesitylene

139.8

18.0

18.0

0

0.6

Halo hydrocarbons Chloromethane

55.4

19.8

15.3

6.1

3.9

Dichloromethane

63.9

20.3

18.2

6.3

7.8

Trichloromethane

80.7

18.8

17.6

3.0

4.2

n-Propyl chloride

88.1

17.4

16.0

7.8

2.0

1,1-Dichloroethene

79.0

18.6

17.0

6.8

4.5

1-Chlorobutane

104.5

17.2

16.2

5.5

2.0

1,2-Dichloroethane

79.4

20.0

19.0

7.4

4.1

Carbon tetrachloride

97.1

17.6

17.6

0

0


107

V1, Kmol/m3

δ, (MJ/m3)1/2

δd, (MJ/m3)1/2

δp, (MJ/m3)1/2

δh, (MJ/m3)1/2

Perchloroethylene

101.1

19.0

19.0 18.8

6.5 0

2.9 1.4

1,1,2,2-Tetrachloroethane

105.2

19.8

18.8

5.1

9.4 5.3

Chloro-difluoromethane (Freon 21)

72.9

17.0

12.3

6.3

5.7

1,1,2-Trichloro-trifluoroethane (Freon 113)

119.2

14.8

14.5

1.6

0

Dichloro-difluoromethane (Freon 12)

92.3

12.2

12.2

2.0

0

Chlorobenzene

102.1

19.5

18.9

4.3

2.0

o-Dichlorobenzene

112.8

20.4

19.1

6.3

3.3

Bromoethane

76.9

19.6

15.8

3.1

5.7

Bromobenzene

105.3

20.3

20.5 18.9

5.5 4.5

4.1 5.1

Solvent

Ethers Epichlorohydrin

72.5

18.5

17.8

1.8

5.3

Tetrahydrofuran

79.9

22.5

19.0

10.2

3.7

1,4-Dioxane

81.7

18.5

16.8

5.7

8.0

Diethyl ether

85.7

20.5

19.0

1.8

7.4

Diisopropyl ether

104.8

15.6

14.4

2.9

5.1

Ketones Acetone

74.0

19.9

15.5

10.4

6.9

Methyl ethyl ketone

90.1

18.9

15.9

9.0 8.4

5.1

Cyclohexanone

104.0

18.8

16.3

7.1

6.1

Diethyl ketone

106.4

18.1

15.8

7.6

4.7

Mesityl oxide

115.6

16.7

15.9 17.6

3.7

4.1

Acetophenone

117.4

19.8

16.5

8.2

7.3

Methyl isobutyl ketone

125.8

17.5

15.3

6.1

4.1

Methyl isoamyl ketone

142.8

17.4

15.9

5.7

4.1

Isophorone

150.5

18.6

16.6

8.2

7.4

Diisobutyl ketone

177.1

16.0

16.0

3.7

4.1

108


Solvent

V1, Kmol/m3

δ, (MJ/m3)1/2

δd, (MJ/m3)1/2

δp, (MJ/m3)1/2

δh, (MJ/m3)1/2

Aldehydes Acetaldehyde

57.1

21.1

14.7

8.0

11.3

Furfural

83.2

22.9

18.6

14.9

5.1

n-Butyraldehyde

88.5

18.4

14.7 18.7

5.3 8.6

7.0

Benzaldehyde

101.5

19.2

19.4

7.4

5.3

Esters Ethylene carbonate

66.0

30.1

29.6

19.4

21.7

Methyl acetate

79.7

19.6

15.5

7.2

7.6

Ethyl formate

80.2

19.6

15.5

8.4

8.4

Propylene-1,2-carbonate

85.0

27.2

20.0

18.0

4.1

n-Propyl formate

97.2

19.5

15.0

5.3

11.2

Propyl acetate

115.1

17.8

15.5

4.5

7.6

Ethyl acetate

98.5

18.6

15.8

5.3

7.2

n-Butyl acetate

132.5

17.4

15.8

3.7

6.3

n-Amyl acetate

149.4

17.3

15.6

3.3

6.7

Isobutyl acetate

133.5

17.0

15.1

3.7

6.3

Isopropyl acetate

117.1

17.3

14.4

6.1

7.4

Diethyl malonate

151.8

19.5

15.5

4.7

10.8

Diethyl oxalate

135.4

22.5

15.5

5.1

15.5

Isoamyl acetate

148.8

16.0

15.3

3.1

7.0

Dimethyl phthalate

163

21.9

18.6

10.8

4.9

Diethyl phthalate

198

20.5

17.6

9.6

4.5

Dibutyl phthalate

266

19.0

17.8

8.6

4.1

Dioctyl phthalate

377

16.8

16.6

7.0

3.1

Phosphorous compounds Trimethyl phosphate

116.7

25.2

16.7

15.9

10.2

Triethyl phosphate

169.7

22.2

16.7

11.4

9.2

316

23.1

19.0

12.3

4.5

Tricresyl phosphate

Nitrogen compounds Acetonitrile

52.6

24.3

15.3

17.9

6.1

n-Butyronitrile

86.7

20.4

15.3

12.5

5.1


109

V1, Kmol/m3

δ, (MJ/m3)1/2

δd, (MJ/m3)1/2

δp, (MJ/m3)1/2

δh, (MJ/m3)1/2

Propionitrile

70.9

22.1

15.3

14.3

5.5

Benzonitrile

102.6

19.9

17.4

9.0

3.3

Nitromethane

54.3

25.1

15.7

18.8

5.1

Nitroethane

71.5

22.6

15.9

15.9

4.5

2-Nitropropane

86.9

20.4

16.1

12.0

4.1

Nitrobenzene

102.7

20.5

20.0

8.6

4.1

Ethylenediamine

67.3

25.2

16.6

8.8

17.0

2-Pyrrolidinone

76.4

30.1

19.4

17.4

11.3

Pyridine

80.9

21.9

19.0

8.8

5.9

Morpholine

87.1

22.1

18.8

4.9

9.2

Aniline

91.5

21.1

19.4

5.1

10.2

n-Butylamine

99.0

17.8

16.2

4.5

8.0

2-Aminoethanol

59.7

31.3

17.1

15.5

21.2

Di-n-propyl amine

136.8

15.9

15.3

1.4

4.1

Diethylamine

103.2

16.4

14.9

2.3

6.1

Quinoline

118.0

22.1

19.4

7.0

7.6

Formamide

39.8

39.3

17.2

26.2

19.0

N,N-Dimethylformamide

77.0

24.8

17.4

13.7

11.2

Solvent

Sulfur compounds Carbon disulfide

60.0

20.3

20.3

0

0

Dimethyl sulfoxide

71.3

26.4

18.4

16.3

10.2

Diethyl sulfide

107.6

17.2

16.8

3.1

2.0

75

29.7

19.0

19.4

12.3

Dimethyl sulfone

Monohydric alcohols and phenols Methanol

40.7

29.1

15.1

12.2

22.2

Ethanol

66.8

26.4

15.8

8.8

19.4

Allyl alcohol

68.4

24.1

16.2

10.8

16.8

1-Propanol

75.2

24.4

15.8

6.7

17.3

2-Propanol

76.8

23.5

15.8

6.1

16.4

Furfuryl alcohol

86.5

25.6

17.4

7.6

15.1

1-Butanol

91.5

23.1

15.9

5.7

15.7

2-Butanol

92.0

22.1

15.8

5.7

14.5

110


V1, Kmol/m3

δ, (MJ/m3)1/2

δd, (MJ/m3)1/2

δp, (MJ/m3)1/2

δh, (MJ/m3)1/2

1-Pentanol

108.3

21.6

15.9

4.5

13.9

Benzyl alcohol

103.6

24.8

18.4

6.3

13.7

Cyclohexanol

106.0

23.3

17.3

4.1

13.5

Ethylene glycol monomethyl ether

79.1

23.3

16.2

9.2

16.4

Ethylene glycol monoethyl ether

97.8

21.5

16.2

9.2

14.3

Ethylene glycol monobutyl ether

142.1

20.8

15.9

4.5

12.7

1-Octanol

157.7

21.1

17.0

3.3

11.9

m-Cresol

104.7

22.7

18.0

5.1

12.9

Solvent

Carboxylic acids Formic acid

37.8

24.8

14.3

11.9

16.6

Acetic acid

57.1

20.7

14.5

8.0

13.5

n-Butyric acid

110

21.5

14.9

4.1

10.6

Polyhydric alcohols Ethylene glycol

55.8

33.2

16.8

11.0

25.9

Glycerol

73.3

43.8

17.3

12.0

29.2

Diethylene glycol

95.3

29.8

16.0

14.7

20.4

Triethylene glycol

114.0

21.9

16.0

12.5

18.6

Water

18.0

47.9

15.5

16.0

42.3

4.1.3 MULTI-DIMENSIONAL APPROACHES These approaches can be divided into three types: 1 H-bonds are not considered. This approach can be applied only for nonpolar and weak polar liquids. 2 H-bonds taken into account by one parameter. 3 H-bonds taken into account by two parameters. Blanks and Prausnitz8,9 proposed two-component solubility parameters. They decomposed the cohesion energy into two contributions of polar and non-polar components: −

E nonpolar E polar E =− − = λ2 + τ 2 V1 V1 V1

where: λ τ

non-polar contribution to solubility parameter polar contribution to solubility parameter

[4.1.15]


111

This approach has become a constituent of the Hansen approach and has not received a separate development. Polar interactions can themselves be divided into two types: • Polar interactions where molecules having permanent dipole moments interact in solution with the dipole orientation in a symmetrical manner. It follows that the geometric mean rule is obeyed for orientation interactions and the contribution of dipole orientations to the cohesive energy and dispersion interactions. • Polar interactions accompanied by the dipole induction. These interactions are asymmetrical. Thus for a pure polar liquid without hydrogen bonds:10 2 δ 2 = δ d2 + δ or + 2δ d δ in

[4.1.16]

where: δd δor δin

dispersion contribution to the solubility parameter orientation contribution to the solubility parameter induction contribution to the solubility parameter.

A more traditional approach of contribution of the induction interaction was published elsewhere;11 however, it was used only for the estimation of the common value of the δ parameter rather than for evaluation of solubility: δ 2 = δ d2 + δ p2 + δ i2

[4.1.17]

The first method taking into account the hydrogen bonding was proposed by Beerbower et al.,12 who expressed hydrogen bonding energy through the hydrogen bonding number ∆ν. The data for various solvents were plotted into a diagram with the solubility parameter along the horizontal axis and the hydrogen bonding number ∆ν along the vertical axis. Data were obtained for a given polymer for suitable solvents. All solvents in which a given polymer was soluble got a certain regions. Lieberman also plotted two-dimensional graphs of solubility parameters versus hydrogen-bonding capabilities.13 On the base of work by Gordy and Stanford, the spectroscopic criterion, related to the extent of the shift to lower frequencies of the OD infrared absorption of deuterated methanol, was selected. It provides a measure of the hydrogen-bonding acceptor power of a solvent.14,15 The spectrum of a deuterated methanol solution in the test solution was compared with that of a solution in benzene and the hydrogen-bonding parameter was defined as γ = ∆ν / 10

[4.1.18]

where: ∆ν

OD absorption shift (in wavenumber).

Crowley et al.16 used an extension of this method by including the dipole moment of the solvents. One of the axis represented solubility parameter, the second the dipole moment, and the third hydrogen bonding expressed by spectroscopic parameter γ. Because this method involved an empirical comparison of a number of solvents it was impractical. Nelson et al.17 utilized this approach to hydrogen bond solubility parameters. Hansen (see the next section) developed this method. Chen introduced a quantity χ H 18

112


χH =

VS RT

[(δ

(

− δ d . P ) + δ p.S − δ p. P 2

d .S

)

2

]

[4.1.19]

where δ d. S , δ d.P , δp. S , δp.P are Hansen’s parameters of solvent and polymer (see the next section). Chen implied that χH was the enthalpy contribution to the Flory-Huggins parameter χ1 and plotted the solubility data in a δ h − χ H diagram where δh was the H-bond parameter in the Hansen approach. In these diagrams sphere-like volumes of Hansen’s solubility have degenerated to circles. The disadvantage of this method lies in the beforehand estimating characteristics of the polymer. Among other two-dimensional methods used for the representation of solubility data was the δp-δh diagram proposed by Henry19 and the δ-δh diagram proposed by Hoernschemeyer,20 but their representations of the solubility region were less correct. All these approaches involving hydrogen bond parameter ignored the fact that hydrogen bond interaction was the product of hydrogen bonding donating and accepting capability.21-23 On the basis of chemical approach to hydrogen bonding, Rider proposed a model of solubility for liquids in which the enthalpy limited the miscibility of polymers and solvents.24,25 For substances capable to form hydrogen bonds, Rider proposed a new factor relating miscibility with an enthalpy of mixing which depends on an enthalpy of the hydrogen bond formation. He has introduced the quantity of a hydrogen bond potential (HBP). If the quantity of HBP is positive it promotes miscibility and if it is negative it decreases miscibility. HBP = (b 1 − b 2 )(C1 − C2 )

[4.1.20]

where: b1,b2 C1, C2

donor parameters of solvent and solute, respectively acceptor parameters of solvent and solute, respectively

For certain polymers Rider has drawn solubility maps. Thus the area of solubility was represented by a pair of symmetric quarters of a plane lying in coordinates b,C.24 Values of parameters were defined from data for enthalpies of hydrogen bonds available from the earlier works. The model is a logical development of the Hansen method. A shortcoming of this model is in neglecting all other factors influencing solubility, namely dispersion and polar interactions, change of entropy, molecular mass of polymer and its phase condition. The model was developed as a three-dimensional dualistic model (see Section 4.1.5). 4.1.4 HANSEN’S SOLUBILITY The Hansen approach26-30 assumed that the cohesive energy can be divided into contributions of dispersion interactions, polar interactions, and hydrogen bonding. E = Ed + Ep + Eh

[4.1.21]

where:

E total cohesive energy Ed, Ep, Eh contributions of dispersion forces, permanent dipole-permanent dipole forces, and hydrogen bonds.

Dividing this equation by the molar volume of solvent, V1, gives:


113

E Ed Ep Eh = + + V1 V1 V1 V1

[4.1.22]

or δ 2 = δ d2 + δ p2 + δ h2

[4.1.23]

where: δ δd, δp, δh

total solubility parameter components of the solubility parameter determined by the corresponding contributions to the cohesive energy.

Hansen gave a visual interpretation of his method by means of three-dimensional spheres of solubility, where the center of the sphere has coordinates corresponding to the values of components of solubility parameter of polymer. The sphere can be coupled with a radius to characterize a polymer. All good solvents for particular polymer (each solvent has been represented as a point in a three-dimensional space with coordinates) should be inside the sphere, whereas all non-solvents should be outside the solubility sphere. An example is given in Section 4.1.7. In the original work these parameters were evaluated by experimental observations of solubility. It was assumed that if each of the solubility parameter components of one liquid is close to the corresponding values of another liquid, then the process of their mixing should readily occur with a more negative free energy. The solubility volume has dimensions δ d, δ p, 2δ h. The factor 2 was proposed to account for the spherical form of solubility volumes and had no physical sense. However, it is necessary to notice that, for example, Lee and Lee31 have evaluated spherical solubility volume of polyimide with good results without using the factor 2. Because of its simplicity, the method has become very popular. Using the Hansen approach, the solubility of any polymer in solvents (with known Hansen’s parameters of polymer and solvents) can be predicted. The determination of polymer parameters requires evaluation of solubility in a great number of solvents with known values of Hansen parameters. Arbitrary criteria of determination are used because Hansen made no attempts of precise calculations of thermodynamic parameters. The separation of the cohesion energy into contributions of various forces implies that it is possible to substitute energy for parameter and sum contributions proportional to the second power of a difference of corresponding components. Hansen’s treatment permits evaluation of the dispersion and polar contribution to cohesive energy. The fitting parameter of the approach (the solubility sphere radius) reflects on the supermolecular structure of polymer-solvent system. Its values should be higher for amorphous polymers and lower for glass or crystalline polymers. The weak point of the approach is the incorrect assignment of the hydrogen bond contribution in the energy exchange that does not permit its use for polymers forming strong hydrogen bonds. Table 4.1.2. Solubility parameters and their components for solvents (after refs 37,40) Polymer Polyamide-66

δ, (MJ/m3)1/2

δd, (MJ/m3)1/2

δp, (MJ/m3)1/2

δh, (MJ/m3)1/2

22.77

18.5

5.1

12.2

114


δ, (MJ/m3)1/2

δd, (MJ/m3)1/2

δp, (MJ/m3)1/2

δh, (MJ/m3)1/2

Polyacrylonitrile

25.10

18.19

15.93

6.74

Polyvinylchloride

21.41

18.68

10.01

3.06

Polymethylmethacrylate

20.18

17.72

5.72

7.76

Polystyrene

19.81

19.68

0.86

2.04

Polytetrafluoroethylene

13.97

13.97

0.00

0

Polyethyleneterephthalate

21.6

19.5

3.47

8.58

Polymer

A large number of data were accumulated for different solvents and polymers (see Tables 4.1.1, 4.1.2). A variation of the Hansen method is the approach of Teas.33 He showed for some polymer-solvent systems that it was possible to use fractional cohesive energy densities plotted on a triangular chart to represent solubility limits: Ed =

δ 2d δ

2 0

, Ep =

δ2p δ

2 0

, Eh =

δ 2h

[4.1.24]

δ 20

where δ 20 = δ 2d + δ 2p + δ 2h Teas used fractional parameters defined as fd =

100δ p 100δ d 100 , fp = , fh = δd + δp + δh δd + δp + δh δd + δp + δh

[4.1.25]

This representation was completely empirical without any theoretical justification. Some correlations between components of solubility parameters and physical parameters of liquids (surface tension, dipole moment, the refraction index) were generalized elsewhere.11 δ 2d + 0.632δ 2p + 0.632δ 2h = 13.9V1−1/ 3 γ l

non-alcohols

[4.1.26]

δ 2d + δ 2p + 0.06δ 2h = 13.9V1−1/ 3 γ l

alcohols

[4.1.27]

δ 2d + 2δ 2p + 0.48δ 2h = 13.9V1−1/ 3 γ l

acids, phenols

[4.1.28]

where: γl

surface tension.

Koenhan and Smolder proposed the following equation applicable to the majority of solvents, except cyclic compounds, acetonitrile, carboxylic acids, and multi-functional alcohols.34 δ 2d + δ 2p = 13.8V1−1/ 3 γ l

[4.1.29]

They also proposed a correlation between polar contribution to the solubility parameter and refractive index:


115

δ d = 9.55n D − 5.55

[4.1.30]

where: nD

refractive index

Alternatively Keller et. al.35 estimated that for nonpolar and slightly polar liquids δ d = 62.8x

for x ≤ 0.28

[4.1.31]

δ d = −4.58 + 108x − 119x 2 + 45x 3

for x > 0.28

[4.1.32]

where:

x=

n D2 − 1 n D2 + 2

Peiffer suggested the following expression:35

(

δ 2d = K 4ππI 2 / 3d

)(N / V )

3

1

[4.1.33]

where: K I α N V1 r*

packing parameter ionization potential molecular polarizability number of molecules in the volume unit =Nr*3/K the equilibrium distance between molecules.

For the estimation of nonpolar component of δ Brown et al.36 proposed the homomorph concept. The homomorph of a polar molecule is the nonpolar molecule most closely resembling it in the size and the structure (e.g., n-butane is the homomorph of n-butyl alcohol). The nonpolar component of the cohesion energy of a polar solvent is taken as the experimentally determined total vaporization energy of the corresponding homomorph at the same reduced temperature (the actual temperature divided by the critical temperature in Kelvin’s scale). For this comparison the molar volumes must also be equal. Blanks and Prausnitz proposed plots of dependencies of dispersion energy density on a molar volume for straight-chain, alicyclic and aromatic hydrocarbons. If the vaporization energies of appropriate hydrocarbons are not known they can be calculated by one of the methods of group contributions (See Chapter 5). Hansen and Scaarup28 calculated the polar component of solubility parameter using Bottcher’s relation to estimating the contribution of the permanent dipoles to the cohesion energy: δ 2p =

12108 ε − 1 n D2 + 2 µ 2 2 2 V1 2ε − n D

(

where: ε µ

dielectric constant, dipole moment

)

[4.1.34]

116


µ δ p = 501 . 3/4 V1

[4.1.35]

Peiffer11 proposed the expressions which separates the contributions of polar forces and induction interactions to the solubility parameters:

(

δ 2p = K 2πµ 4 / 3kTp

) (N / V )

δ 2i = K (2παµ / i )(N / V1 )

3

[4.1.36]

1

3

[4.1.37]

where: ε* N

interaction energy between two molecules at the distance r* number of hydroxyl groups in molecule

p = 2µ 4 / 3kTε *

[4.1.38]

i = 2αµ 2 / ε *

[4.1.39]

It should be noted that in Hansen’s approach these contributions are cumulative: δ 2p = δ 2p + δ 2i

[4.1.40]

For the calculation of hydrogen-bonding component, δh, Hansen and Scaarup28 proposed an empirical expression based on OH-O bond energy (5000 cal/mol) applicable to alcohols only: δ h = (20.9N / V1 )

1/ 2

[4.1.41]

In Subchapter 5.3, the values of all the components of a solubility parameter are calculated using group contributions. 4.1.5 THREE-DIMENSIONAL DUALISTIC MODEL The heat of mixing of two liquids is expressed by the classical theory of regular polymer solutions using Eq. [4.1.10]. This expression is not adequate for systems with specific interactions. Such interactions are expressed as a product of the donor parameter and the acceptor parameter. The contribution of H-bonding to the enthalpy of mixing can be written in terms of volume units as follows:21 ∆H′mix = (A1 − A2 )(D1 − D2 ) ϕ1 ϕ 2

[4.1.42]

where: A1, A2 D1, D2 ϕ 1, ϕ 2

effective acceptor parameters donor parameters, volume fractions,

Hence enthalpy of mixing of two liquids per volume unit can be expressed by:32 ∆Hmix =

[(δ′ − δ′ ) 1

2

2

]

+ (A1 − A2 )(D1 − D2 ) ϕ1 ϕ 2 = Bϕ1 ϕ 2

[4.1.43]


117

This equation is used for the calculation of the enthalpy contribution to the Huggins parameter (see Subchapter 4.2) for a polymer-solvent system: χ H = V1

[(δ′ − δ′ ) 1

2

2

]

+ (A1 − A2 )(D1 − D2 ) / RT

[4.1.44]

where: δ1′, δ 2′

dispersion-polar components of solubility parameters (values of solubility parameters excluding H-bonds contributions).

Results of calculations using Eq. [4.1.44] of three-dimension dualistic model coincide with the experimental values of χH and the χH values calculated by other methods.37 Values A, D, and δ′ can been obtained from IR-spectroscopy evaluations and Hansen’s parameters.24,25 Values of the TDM parameters are presented in Tables 4.1.3, 4.1.4. It should be noted that Hansen parameters are used for estimation of values of TDM parameters from equation δ 2di + δ 2pi = δ′i 2 , δ hi2 = A i Di . Table 4.1.3. TDM parameters of some solvents. [Adapted, by permission, from V.Yu. Senichev, V.V. Tereshatov , Vysokomol . Soed ., B31 , 216 (1989).]

#

Solvent

D

A

δ′

δ

V1×106 m3/mol

3 1/2

(MJ/m )

1

Isopropanol

11.8

13.3

20.0

23.6

76.8

2

Pentanol

9.9

11.2

19.7

22.3

108.2

3

Acetone

3.8

13.1

18.5

19.8

74.0

4

Ethyl acetate

4.9

10.4

17.0

18.3

98.5

5

Butyl acetate

4.8

9.0

16.4

17.6

132.5

6

Isobutyl acetate

4.8

8.4

15.7

16.9

133.3

7

Amyl acetate

4.7

7.8

16.2

17.3

148.9

8

Isobutyl isobutyrate

4.7

7.4

14.6

15.7

165.0

9

Tetrahydrofuran

5.2

12.2

17.7

19.4

81.7

10

o-Xylene

0.5

7.4

18.3

18.4

121.0

11

Chlorobenzene

0.6

7.3

19.3

19.4

102.1

12

Acetonitrile

2.7

14.0

24.4

24.5

52.6

13

n-Hexane

0

0

14.9

14.9

132.0

14

Benzene

0.6

8.5

18.6

18.8

89.4

15


8.2

15.4

22.0

27.7

77.0

16

Toluene

0.6

7.8

18.1

18.2

106.8

17

Methanol

18.2

17.0

23.5

29.3

41.7

18

Ethanol

14.4

14.7

21.7

26.1

58.5

118


#

D

Solvent

A

δ′

δ

V1×106 m3/mol

3 1/2

(MJ/m )

19

1-Propanol

11.9

13.4

21.0

24.5

75.2

20

Methyl ethyl ketone

2.3

11.6

18.3

19.0

90.1

21

Cyclohexanone

2.7

11.1

19.5

20.3

104.0

22

Diethyl ether

7.6

12.4

12.0

15.4

104.8

23

Ethylbenzene

0.3

7.3

17.9

18.0

123.1

24

Pyridine

1.9

19.6

21.0

21.8

80.9

25

Propyl acetate

4.8

9.6

16.4

17.8

115.2

26

1,4-Dioxane

8.3

12.9

17.7

20.5

85.7

27

Aniline

6.2

16.8

20.0

22.5

91.5

Table 4.1.4. TDM parameters of some polymers, (MJ/m3)1/2. [Adapted, by permission, from V.Yu. Senichev, V.V. Tereshatov , Vysokomol . Soed ., B31 , 216 (1989)] D

A

δ′

δ


2.5

6.5

18.6

19.8

Polyvinylacetate

4.9

10.4

17.8

19.2

Polystyrene

0.3

7.3

17.9

18.0

Polyvinylchloride

11.6

10.6

16.1

19.5

Polymer

δh can be separated into donor and acceptor components using values of enthalpies of the hydrogen bond formation between proton-donors and proton-acceptors. In the absence of such data it is possible to evaluate TDM parameters by means of analysis of parameters of compounds similar in the chemical structure. For example, propyl acetate parameters can been calculated by the interpolation of corresponding parameters of butyl acetate and ethyl acetate,24 parameters of benzene can be calculated by decomposition of δh into acceptor and donor components in the way used for toluene elsewhere.25 The solubility prediction can be made using the relationship between solubility and the χ 1 parameter (see Subchapter 4.2). The total value of the χ 1 parameter can be evaluated by adding the entropy contribution: χ1 = χS + χ H

[4.1.45]

where: χS

an empirical value. Usually it is 0.2-0.4 for good solvents.

The value of the parameter is inversely proportional to coordination number that is number of molecules of a solvent interacting with a segment of polymer. The value of the entropy contribution to the parameter should be included in solubility calculations. The value χ S = 0.34 is then used in approximate calculations.


119

4.1.6 SOLUBILITY CRITERION The polymer superstructure influences its solubility. Askadskii and Matveev proposed a new criterion of solubility for linear polymers based on interaction of forces of a surface tension on wetting.38 The solubility parameter of polymer should be lower or equal to the work of rupture by solvent of a bond relative to a volume unit of the bond element. The condition of solubility can be expressed as follows:  γp µ ≤ 2ρΦ  γs

1/ 2

   

[4.1.46]

where: µ δp, δs

= δ 2p / δ 2s solubility parameters for polymer and solvent accordingly.

p εmax rs

ρ=

[4.1.47]

εsmax rp

Φ=

4(VpVs )

(V

1/ 2 p

1/ 2

+ Vs1/ 2

)

[4.1.48]

2

where: s ε pmax , ε max rs, rp

Vp, Vs Φ ≈1 ρ ≈ const

maximum deformations of polymer and solvent at rupture characteristic sizes of Frenkel’s swarms for solvent and and small radius of globule of bond for polymer, respectively molar volumes of polymer and solvent (per unit)

The above expression was obtained with neglecting the preliminary swelling. Consideration of swelling requires correction for surface tension of swelled surface layers:

(

µ < 2ρΦ Φ − Φ 2 − 1 + a 

)

1/ 2

 

[4.1.49]

where:

a = γ ps / γ p

[4.1.50]

γ ps = γ p + γ s − 2Φ( γ p γ s )

1/ 2

[4.1.51]

For practical purposes, the magnitude of ρ estimated graphically is 0.687. Thus µ < 1374 . β

[4.1.52]

(

for β = Φ Φ − Φ 2 − 1 + a 

)

1/ 2

 

120


For both polymers and solvents, the values of solubility parameters can be obtained experimentally (see Subchapters 5.1, 5.3). The surface tension of polymer can be calculated using parahor: γ = (P / V)

4

[4.1.53]

where: V

molar volume of a repeated polymer unit

Then the value of Vp is calculated: Vp =

NA ∑ ∆Vi i

[4.1.54]

k av

where: kav

= 0.681

If the density of polymer dp is known, then Vp = M/dp, where M is the molecular mass of a repeating unit. The values of parahors are given in Table 4.1.5. Table 4.1.5. Values of parahors Atom

C

H

O

O2

N

S

F

Cl

Br

I

P

4.8

17.1

20.0

60.0

12.5

48.2

27.5

54.3

68.0

91.0

Increment

Double bond

Triple bond

3-member ring

4-member ring

5-member ring

6-member ring

P

23.2

46.4

16.7

11.6

8.5

6.1

The value of Φ is calculated from Eq. [4.1.48]. Vp, Vs are defined from ratios Vp=M/dp and Vs=M/ds where dp, ds are the densities of polymer and solvent, respectively. Then µ is calculated from Eq. [4.1.49]. The obtained value of µ from Eq. [4.1.49] is compared with value of µ = δ 2p / δ 2s if the last value is lower or equal to the value of µ calculated from Eq. [4.1.49], polymer should dissolve in a given solvent with probability of 85 %. 4.1.7 SOLVENT SYSTEM DESIGN One-component system. Solvents can be arranged in accordance to their solubility parameter as shown in Figure 4.1.1. It is apparent that a set of compatible solvents can be selected for polymer, determining their range based on the properties of polymer. The simplest case is expressed by the Gee’s equation for equilibrium swelling:41

[

Q = q max exp −Vs ( δs − δ p )

2

]

[4.1.55]

where: δs, δp

solubility parameters for solvent and polymer.

The value of solubility parameter of solvent mixture with components having similar molar volumes is relative to their volume fractions and solubility parameters:


Figure 4.1.1. Representation of the one-component solubility parameter system. The curve represents Eq. [4.1.56] for polymer with δ =18 (MJ/m3)1/2.

Figure 4.1.3. The solubility volume for cellulose acetate butyrate in terms of one-component solubility parameter, δ, dipole moment, µ, and (on vertical axis) the spectroscopic parameter, γ, from the approach developed by Crowley et.al.2 [Adapted, by permission, from J.D. Crowley, G.S. Teague and J.W. Lowe, J. Paint Technol., 38, 269 (1966)]

δ = ∑ δi ϕ i i

[4.1.56]

121

Figure 4.1.2. Representation of the two-component solubility parameter system in the Rider’s approach. b2, C2 are the values of Rider’s parameters for polymer.

Figure 4.1.4. Hansen’s solubility volume of polyimide synthesized from 3,3’4,4’-benzophenone tetracarboxylic dianhydride and 2,3,5,6-tetramethyl-p-phenylene diamine (after Lee31). [Adapted, by permission, from H.-R. Lee, Y.-D. Lee, J. Appl. Polym. Sci., 40, 2087 (1990)]

Solute is frequently soluble in a mixture of two non-solvents, for example, the mixture of diisopropyl ether (δ = 15.6 (MJ/m3)1/2) and ethanol (δ = 26.4 (MJ/m3)1/2) is a solvent for nitrocellulose (δ = 23 (MJ/m3)1/2). Two-component systems. Two parametrical models of solubility use two-dimensional graphs of solubility area. Two-dimensional solubility areas may be closed or open.

122


In accordance to the approach of Blanks and Prausnitz, the solubility area is displayed as a plane with two coordinates, λ, τ.8,9 The parameters are related to Hansen’s parameters by λ = δd ,

(

τ = δ p2 + δ h2

)

1/ 2

Solubility areas in this approach are closed because they are degenerated from Hansen’s spheres (see below). Another example of apFigure 4.1.5. Volume of the increased swelling of crosslinked plication of degenerate polybutadiene urethane elastomer, χ H ≤ 0.85. Labels of points: 1-outside volume, points with coordinates corresponding to solvents; 2-in- Hansen’s spheres was given side volume, points with coordinates corresponding to solvents; 3-the by Chen.18 Instead of paramepoints with coordinates corresponding to polymer. This is the center of the ters δp, δd the value χH is used volume. 4- points placed on a plane with coordinate δ =18 (MJ/m3)1/2. which is calculated from the Number of solvent (not underlined number) corresponds to their position in the Table 4.1.3. The underlined number corresponds to swelling ratio at difference of the polar and the equilibrium. dispersing contributions, δp, δd. The zone of solubility is a circle. Lieberman13 uses planes with coordinates δ, γ where γ is spectroscopic parameter (see Section 4.1.3). These planes are open and have the areas of solubility, non-solubility and intermediate. In Rider’s approach, the solubility area is a system of two quarters on a plane; two other quarters are the areas of non-solubility (Figure 4.1.2). Coordinates of this plane are accepting and donating abilities. Rider’s approach finds application for solvents with high H-bond interactions. Three-component systems. Crowley et. al.2 proposed the three-dimensional solubility volumes (Figure 4.1.3). Better known are Hansen’s three-dimensional solubility volumes (Figure 4.1.4). In Hansen’s approach, the components of solubility parameters for mixed solvents δj are calculated from Eq. [4.1.56]: δ j = ∑ δ ji ϕ i

[4.1.57]

i

The choice of a solvent for polymer is based on coordinates of polymer in space of coordinates δd, δp, δh and the radius of a solubility volume. For mixed solvents, their coordinates can be derived by connecting coordinates of individual solvents. This can be used to determine synergism in solvents for a particular polymer but it cannot demonstrate antisynergism of solvent mixtures (a mixture is less compatible with polymer than the individual solvents).


123

In the TDM approach, the solubility volume has also three coordinates, but the H-bond properties are taken into account. For this reason solubility volume can be represented by hyperbolic paraboloid (Figure 4.1.5). The use of this model permits to evaluate potential of synergism and antisynergism of solvent mixtures. It can be demonstrated by position of a point of a solvent mixture moving from a zone of good compatibility into the similar zone through the zone of inferior compatibility or, on the contrary, from a zone of an incompatibility into a similar zone through the zone of the improved compatibility. In the case of polymers that have no hydrogen bond forming abilities, this approach is equivalent to the Hansen or Blanks-Prausnitz approaches. The above review of the methods of solvent evaluation shows that there is a broad choice of various techniques. Depending on the complexity of solvent-polymer interactions, the suitable method can be selected. For example, if solvents and/or polymer do not have functional groups to interact with as simple method as one-dimensional model is adequate. If weak hydrogen bonding is present, Hansen’s approach gives good results (see further applications in Subchapter 5.3). In even more complex interactions, TDM model is always the best choice. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

W.W. Reynolds and E.C. Larson, Off. Dig. Fed. Soc. Paint Technol., 34, 677 (1972). J.D. Crowley, G.S.Teague and J.W. Lowe, J. Paint Technol., 38, 269 (1966). H. Burrel, Off. Dig. Fed. Paint. Varn. Prod. Clubs, 27, 726 (1955). P.L. Huyskens, M.C.Haulait-Pirson, Org. Coat. Sci. Technol., 8, 155 (1986). I.H. Hildebrand , J. Amer. Chem. Soc., 38, 1452, (1916). G. Scatchard, J. Amer. Chem. Soc., 56, 995 (1934). A.A. Tager, L.K. Kolmakova, Vysokomol. Soed., A22, 483 (1980). R.F. Blanks, J.M. Prausnitz, Ind. Eng. Chem., Fundam., 3, 1, (1964). R.F. Weimer and J.M. Prausnitz, Hydrocarbon Process., 44, 237 (1965). R.A Keller, B.L.Karger and L.R. Snyder, Gas Chromatogr., Proc. Int Symp. (Eur), 8, 125 (1971). D.G. Peiffer, J. Appl. Polym. Sci., 25, 369 (1980). A. Beerbower, L.A. Kaye and D.E. Pattison, Chem. Eng., 118 (1967). E.P. Lieberman, Off. Dig. Fed. Soc. Paint Technol., 34, 30 (1962). W.Gordy and S.C. Stanford, J. Chem. Phys., 8, 170 (1940). W.Gordy and S.C. Stanford, J. Chem. Phys., 9, 204 (1941). J.D. Crowley, G.S. Teague and J.W. Lowe, J. Paint Technol., 39, 19 (1967). R.C. Nelson, R.W. Hemwall and G.D Edwards, J. Paint Technol., 42, 636 (1970). S.-A. Chen, J. Appl. Polym. Sci., 15, 1247 (1971). L.F. Henry, Polym. Eng. Sci., 14, 167 (1974) D. Hoernschemeyer, J. Appl. Polym. Sci., 18, 61 (1974). P.A. Small, J. Appl. Chem., 3, 71 (1953). H. Burrel, Adv. Chem. Ser., No. 124, 1 (1973). H. Renon, C.A Reckert and J.M. Prausnitz, Ind. Eng. Chem., 3, 1(1964). P.M Rider, J. Appl. Polym. Sci., 25, 2975 (1980). P.M Rider, Polym. Eng. Sci., 23, 180 (1983). C.M. Hansen, J. Paint. Technol., 39, 104 (1967). C.M. Hansen, J. Paint. Technol., 39, 505 (1967). C.M. Hansen and K. Scaarup, J. Paint. Technol., 39, 511 (1967). C.M. Hansen, Three Dimensional Solubility Parameters and Solvent Diffusion Coefficient, Danish Technical Press, Copenhagen, 1967. C.M. Hansen and A. Beerbower in Kirk-Othmer Encyclopedia of Chemical Technology, Suppl. Vol., 2nd ed., A.Standen Ed., 889, 1971. H.-R. Lee, Y.-D. Lee, J. Appl. Polym.Sci., 40, 2087 (1990). V.Yu. Senichev, V.V. Tereshatov, Vysokomol. Soed., B31, 216 (1989). J.P.Teas, J. Paint Technol., 40, 19 (1968). D.M. Koenhen, C.A. Smolder, J. Appl. Polym. Sci., 19, 1163 (1975).

124

35 36 37 38 39 40 41

Valery Yu. Senichev, Vasiliy V. Tereshatov R.A Keller, B.L. Karger and L.R. Snyder, Gas Chromtogr., Proc. Int Symp. (Eur), 8, 125(1971). H.C. Brown, G.K. Barbaras, H.L. Berneis, W.H. Bonner, R.B. Johannesen, M. Grayson and K.L. Nelson, J. Amer. Chem. Soc., 75, 1 (1953). A.E. Nesterov, Handbook on physical chemistry of polymers. V. 1. Properties of solutions, Naukova Dumka, Kiev, 1984. A.A. Askadskii, Yu.I. Matveev, M.S. Matevosyan, Vysokomol. Soed. 32, 2157 (1990). Yu.I. Matveev, A.A. Askadskii, Vysokomol. Soed., 36, 436 (1994). S.A. Drinberg, E.F. Itsko, Solvents for paint technology, Khimiya, Leningrad, 1986. A.F. Barton, Chem Rev., 75, 735 (1975).

4.2 EFFECT OF SYSTEM VARIABLES ON SOLUBILITY Valery Yu. Senichev, Vasiliy V. Tereshatov Institute of Technical Chemistry Ural Branch of Russian Academy of Sciences, Perm, Russia

Solubility in solvents depends on various internal and external factors. Chemical structure, molecular mass of solute, and crosslinking of polymer fall into the first group of factors, in addition to temperature and pressure in the second group of factors involved. 4.2.1 GENERAL CONSIDERATIONS The process of dissolution is determined by a combination of enthalpy and entropy factors. The dissolution description can be based on the Flory-Huggins equation. Flory1-3 and Huggins4 calculated the entropy of mixing of long-chain molecules under the assumption that polymer segments occupy sites of a “lattice” and solvent molecules occupy single sites. The theory implies that the entropy of mixing is combinatorial, i.e., it is stipulated by permutations of molecules into solution in which the molecules of mixed components differ greatly in size. The next assumption is that ∆Vmix = 0 and that the enthalpy of mixing does not influence the value of ∆Smix. The last assumptions are the same as in the Hildebrand theory of regular solutions.5 The expression for the Gibbs energy of mixing is  V ∆G = x 1 ln ϕ1 + x 2 ln ϕ 2 + χ 1 ϕ1 ϕ 2  x 1 + x 2 2 RT V1 

  

[4.2.1]

where: x1, x2 χ1

molar fractions of solvent and polymer, respectively Huggins interaction parameter

The first two terms result from the configurational entropy of mixing and are always negative. For ∆G to be negative, the χ1 value must be as small as possible. The theory assumes that the χ1 parameter does not depend on concentration without experimental confirmation. χ1 is a dimensionless quantity characterizing the difference between the interaction energy of solvent molecule immersed in the pure polymer compared with interaction energy in the pure solvent. It is a semi-empirical constant. This parameter was introduced by Flory and Huggins in the equation for solvent activity to extend their theory for athermic processes to the non-athermic processes of mixing:

4.2 Effect of system variables on solubility

ln a1 =

∆µ 1 = ln(1 − ϕ 2 ) + ϕ 2 + χ 1 ϕ 22 RT

125

[4.2.2]

where: * χ 1 = z∆ε12 / kT

[4.2.3]

* * , a1 - solvent activity, ε11 , ε 22 - energy of 1-1 and 2-2 contacts + ε*22 ) − ε12 ∆ε*12 = 0.5( ε11 formation in pure components, ε12 - energy of 1-2 contacts formation in the mixture, µ 1chemical potential of solvent. The critical value of χ1 sufficient for solubility of polymer having large molecular mass is 0.5. Good solvents have a low χ1 value. χ1 is a popular practical solubility criterion and comprehensive compilations of these values have been published.6-9 Temperature is another factor. It defines the difference between polymer and solvent. Solvent is more affected than polymer. This distinction in free volumes is stipulated by different sizes of molecules of polymer and solvent. The solution of polymer in chemically identical solvent should have unequal free volumes of components. It causes important thermodynamic consequences. The most principal among them is the deterioration of compatibility between polymer and solvent at high temperatures leading to phase separation. The theory of regular solutions operates with solutions of spherical molecules. For the long-chain polymer molecules composed of segments, the number of modes of arrangement in a solution lattice differs from a solution of spherical molecules, and hence it follows the reduction in deviations from ideal entropy of mixing. It is clear that the polymer-solvent interactions differ qualitatively because of the presence of segments. Some novel statistical theories of solutions of polymers use the χ1 parameter, too. They predict the dependence of the χ1 parameter on temperature and pressure. According to the Prigogine theory of deformable quasi-lattice, a mixture of a polymer with solvents of different chain length is described by the equation:10

Rχ 1 = A(rA / T ) + (BT / rA )

[4.2.4]

where: A, B rA

constants number of chain segments in homological series of solvents.

These constants can be calculated from heats of mixing, values of parameter χ1, and from swelling ratios. The Prigogine theory was further developed by Patterson, who proposed the following expression:11 U  C  χ1 =  1 ν 2 +  P 1  τ 2 RT    2R 

[4.2.5]

where: U1 CP1 ν, τ

configuration energy (-U1 - enthalpy energy) solvent thermal capacity molecular parameters

The first term of the equation characterizes distinctions in the fields of force of both sizes of segments of polymer and solvent. At high temperatures, in mixtures of chemically similar components, its value is close to zero. The second term describes the structural con-

126


tribution to χ1 stipulated by the difference in free volumes of polymer and solvent. Both terms of the equation are larger than zero, and as temperature increases the first term decreases and the second term increases. The expression can be given in a reduced form (with some additional substitutions):12   ~ 1/ 3 P  V 1  X 12 χ1  = V1* RT1*  ~ 1/ 3  P1* V 1 − 1  * 1

where:

   V1 2 + τ  1/ 3   2 4 / 3 − V~   1      ~ 1/ 3

[4.2.6]

~

V1 , P1* , T1* reduced molar volume of solvent, pressure and temperature consequently contact interaction parameter. X12

These parameters can be calculated if factors of the volumetric expansion, isothermal compressibility, thermal capacity of a solvent and enthalpy of mixing of solution components are known. With temperature decreasing, the first term of the right side of the expression [4.2.6] increases and the second term decreases. Such behavior implies the presence of the upper and lower critical temperatures of mixing. Later Flory developed another expression for χ1 that includes the parameter of contact interactions, X12:13,14 χ1 =

P1*V1*  s 2  ~  V 1 RT  s 1

 X 12 α 1T   P2*  +  * 2   P1*  P1 2

 s X   τ − 2 12   s 1 P1*  

2

   

[4.2.7]

where: s1, s2

ratios of surfaces of molecules to their volumes obtained from structural data.

The large amount of experimental data is then an essential advantage of the Flory’s theory.9 Simple expressions exist for parameter X12 in the terms of Xij characteristic parameters for chemically different segments of molecules of components 1 and 2. Each segment or chemical group has an assigned value of characteristic length (αi, αj) or surface area as a fraction of the total surface of molecule:15

(

)

Χ12 = ∑ (α i,1 − α i, 2 ) α j,1 − α j, 2 Χ ij i, j

[4.2.8]

Bondi’s approach may be used to obtain surface areas of different segments or chemical groups.16 To some extent Huggins’ new theory17-21 is similar to Flory’s theory. 4.2.2 CHEMICAL STRUCTURE Chemical structure and the polarity determine dissolution of polymers. If the bonds in polymer and solvent are similar, then the energy of interaction between homogeneous and heterogeneous molecules is nearly identical which facilitates solubility of polymer. If the chemical structure of polymer and solvent molecule differ greatly in polarity, then swelling and dissolution does not happen. It is reflected in an empirical rule that “like dissolves like”.


127

Nonpolar polymers (polyisoprene, polybutadiene) mix infinitely with alkanes (hexane, octane, etc.) but do not mix with such polar liquids as water and alcohols. Polar polymers (cellulose, polyvinylalcohol, etc.) do not mix with alkanes and readily swell in water. Polymers of the average polarity dissolve only in liquids of average polarity. For example, polystyrene is not dissolved or swollen in water and alkanes but it is dissolved in aromatic hydrocarbons (toluene, benzene, xylene), methyl ethyl ketone and some ethers. Polymethylmethacrylate is not dissolved nor swollen in water nor in alkanes but it is dissolved in dichloroethane. Polychloroprene does not dissolve in water, restrictedly swells in gasoline and dissolves in 1,2-dichloroethane and benzene. Solubility of polyvinylchloride was considered in terms of relationship between the size of a solvent molecule and the distance between polar groups in polymer.22 The above examples are related to the concept of the one-dimensional solubility parameter. However the effects of specific interactions between some functional groups can change compatibility of the system. Chloroalkanes compared with esters are known to be better solvents for polymethylmethacrylate. Aromatic hydrocarbons although having solubility parameters much higher than those of alkanes, dissolve some rubbers at least as well as alkanes. Probably it is related to increase in entropy change of mixing that has a positive effect on solubility. The molecular mass of polymer significantly influences its solubility. With molecular mass of polymer increasing, the energy of interaction between chains also increases. The separation of long chains requires more energy than with short chains. 4.2.3 FLEXIBILITY OF A POLYMER CHAIN The dissolution of polymer is determined by chain flexibility. The mechanism of dissolution consists of separating chains from each other and their transfer into solution. If a chain is flexible, its segments can be separated without a large expenditure of energy. Thus functional groups in polymer chain may interact with solvent molecules. Thermal movement facilitates swelling of polymers with flexible chains. The flexible chain separated from an adjacent chain penetrates easily into solvent and the diffusion occurs at the expense of sequential transition of links. The spontaneous dissolution is accompanied by decrease in free energy (∆G < 0) and that is possible at some defined values of ∆H and ∆S. At the dissolution of high-elasticity polymers ∆H ≥ 0, ∆S > 0 then ∆G < 0. Therefore high-elasticity polymers are dissolved in solvents completely. The rigid chains cannot move gradually because separation of two rigid chains requires large energy. At usual temperatures the value of interaction energy of links between polymer chains and molecules of a solvent is insufficient for full separation of polymer chains. Amorphous linear polymers with rigid chains having polar groups swell in polar liquids but do not dissolve at normal temperatures. For dissolution of such polymers, the interaction between polymer and solvent (polyacrylonitrile in N,N-dimethylformamide) must be stronger. Glassy polymers with a dense molecular structure swell in solvents with the heat absorption ∆H > 0. The value of ∆S is very small. Therefore ∆G > 0 and spontaneous dissolution is not observed and the limited swelling occurs. To a greater degree this concerns crystalline polymers which are dissolved if ∆H < 0 and |∆H| > |T∆S|. When molecular mass of elastic polymers is increased, ∆H does not change but ∆S decreases. The ∆G becomes less negative. In glassy polymers, the increase in molecular mass

128


is accompanied by a decrease in ∆H and ∆S. The ∆S value changes faster than the ∆H value, therefore the ∆G value becomes more negative, which means that the dissolution of polymeric homologues of the higher molecular weight becomes less favorable. Crystalline polymers dissolve usually less readily than amorphous polymers. Dissolution of crystalline polymers requires large expenditures of energy for chain separation. Polyethylene swells in hexane at the room temperature and dissolves at elevated temperature. Isotactic polystyrene does not dissolve at the room temperature in solvents capable to dissolve atactic polystyrene. To be dissolved, isotactic polystyrene must be brought to elevated temperature. 4.2.4 CROSSLINKING The presence of even a small amount of crosslinks hinders chain separation and polymer diffusion into solution. Solvent can penetrate into polymer and cause swelling. The swelling degree depends on crosslink density and compatibility of polymer and solvent. The correlation between thermodynamic parameters and the value of an equilibrium swelling is given by Flory-Rehner equation23 used now in a modified form:24 ln(1 − ϕ 2 ) + ϕ 2 + χ 1 ϕ 22 = −

ν 2  1/ 3 2ϕ 2  Vs  ϕ 2 −  V f  

[4.2.9]

where: ϕ2 ν 2 /V f

polymer volume fraction in a swollen sample volume concentration of elastically active chains the functionality of polymer network

The value of ν 2 /V is determined by the concentration of network knots. These knots usually have a functionality of 3 or 4. This functionality depends on the type of curing agent. Crosslinked polyurethanes cured by polyols with three OH-groups are examples of the three-functional network. Rubbers cured through double bond addition are examples of four-functional networks. Eq. [4.2.9] has different forms depending on the form of elasticity potential but for practical purposes (evaluation of crosslinking density of polymer networks) it is more convenient to use the above form. The equation can be used in a modified form if the concentration dependence of the parameter χ1 is known. The value of equilibrium swelling can be a practical criterion of solubility. Good solubility of linear polymers is expected if the value of equilibrium swelling is of the order of 300-400%. The high resistance of polymers to solvents indicates that the equilibrium swelling does not exceed several percent. In engineering data on swelling obtained at non-equilibrium conditions (for example, for any given time), swelling is frequently linked to the diffusion parameters of a system (see more on this subject in Subchapter 6.1).25 An interesting effect of swelling decrease occurs when swollen polymer is placed in a solution of linear polymer of the same structure as the crosslinked polymer. The decrease of solvent activity causes this effect. The quantitative description of these processes can be made by the scaling approach.26 4.2.5 TEMPERATURE AND PRESSURE The temperature effect on solubility may have different characters depending on the molecular structure of solute. For systems of liquid-amorphous polymer or liquid-liquid, the tem-


Figure 4.2.1 Two contributions to the χ1 parameter. 1-χH, 2 -χS, 3 - the total value of χ1.

129

Figure 4.2.2 The χ1 parameter as a function of pressure at T = 300K, r1 = 3.5. The curves are for the following values of ε11* / ε*22 = 0.85; 1.0 and 1.3 (After refs.9,28).

perature raise can cause improvement of compatibility. Such systems are considered to have the upper critical solution temperature (UCST). If the system of two liquids becomes compatible at any ratio at the temperature below the defined critical point, the system is considered to have the lower critical solution temperature (LCST). Examples of a system with UCST are mixtures of methyl ethyl ketone-water (150oC) and phenol-water (65.8oC). An example of a system with LCST is the mixture of water-triethylamine (18oC). There are systems with both critical points, for example, nicotine-water and glycerol-benzyl-ethylamine. Presence of UCST in some cases means a rupture of hydrogen bonds on heating; however, in many cases, UCST is not determined by specific interactions, especially at high temperatures, and it is close to critical temperature for the liquid-vapor system. There are suppositions that the UCST is the more common case but one of the critical points for polymer-solvent system is observed only at high temperatures. For example, polystyrene (M = 1.1×105) with methylcyclopentane has LCST 475K and UCST 370K. More complete experimental data on the phase diagrams of polymer-solvent systems are published elsewhere.27 The solubility of crystalline substances increases with temperature increasing. The higher the melting temperature of the polymer, the worse its solubility. Substances having higher melting heat are less soluble, with other characteristics being equal. Many crystalline polymers such as polyethylene or polyvinylchloride are insoluble at the room temperature (only some swelling occurs); however, at elevated temperature they dissolve in some solvents. The experimental data on the temperature dependence of χ1 of polymer-solvent systems are described by the dependence χ 1 = α + β / T. Often in temperatures below 100oC, β < 0.9 In a wide temperature range, this dependence decreases non-linearly. The negative contribution to ∆S and positive contribution to the χ1 parameter are connected with the difference of free volume. On heating the difference in free volumes of polymer and solvent in-

130


creases as does the contribution into χS (Figure 4.2.1). For example, polyvinylchloride with dibutyl phthalate, tributylphosphate and some other liquids have values β > 0. The dependence of solubility on pressure can be described only by modern theories taking into account the free volume of components.28 The corresponding states theory predicts12 a pressure dependence of the χ1 parameter through the effect on the free volume of the solution components. This dependence is predicted by the so-called solubility parameters theory as well,28 where the interaction between solvent and solute is described by Figure 4.2.3. A phase diagram with the lower critical solubility parameters with their dependencies solution temperature (LCST).1-binodal, 2-spinodal; and pressure (Fig. 4.2.2). I- zone of non-stable conditions, II- zone of on temperature * * When / ε ε metastable conditions, III- zone of the one phase con11 22 ≤ 1 then δ1 < δ 2 and hence ditions. ∂χ 1 / ∂P < 0in the solubility parameters theory and in the corresponding states theory. When ε*11 / ε*22 is greater than unity δ1 > δ 2 and the solvent becomes less compressible than the polymer. Then pressure can increase the (δ1 − δ2 ) value, giving ∂χ 1 / ∂P > 0. 4.2.6 METHODS OF CALCULATION OF SOLUBILITY BASED ON THERMODYNAMIC PRINCIPLES Within the framework of the general principles of thermodynamics of solutions, the evaluation of solubility implies the evaluation of value of the Gibbs energy of mixing in the whole range of concentrations of solution. However, such evaluation is difficult and for practical purposes frequently unnecessary. The phase diagrams indicate areas of stable solutions. But affinity of solvent to polymer in each of zone of phase diagram differs. It is more convenient to know the value of the interaction parameter, possibly with its concentration dependence. Practical experience from solvent selection for rubbers gives foundations for use of equilibrium swelling of a crosslinked elastomer in a given solvent as a criterion of solubility. The equilibrium swelling is related to χ1 parameter by Eq. [4.2.9]. As previously discussed in Subchapter 4.1, the value of the χ1 parameter can be determined as a sum of entropy and enthalpy contributions. In the one-dimensional solubility parameter approach, one may use the following equation: χ1 = χS +

(δ1 − δ2 )

2

V1

RT

[4.2.10]

where: χS

the entropy contribution

In TDM approach, Eq. [4.1.45] can be used. Similar equations can be derived for the Hansen approach. All existing systems of solubility imply some constancy of the entropy contribution or even constancy in some limits of a change of cohesion characteristics of polymers. Frequently χ1 = 0.34 is used in calculations.


131

Phase diagrams are characterized by critical temperatures, spinodals and binodals. A binodal is a curve connecting equilibrium structures of a stratified system. A spinodal is a curve defining boundary of metastables condition (Fig.4.2.3). Binodals are evaluated experimentally by light scattering at cloud point,29 by volume changes of coexisting phases,30 or by the electron probe R-spectral analysis.31 It is possible to calculate phase behavior, considering that binodals correspond to a condition:

(∆µ i )

′

= (∆µ i )

″

[4.2.11]

where: i

a component of a solution ′

″

changes of a chemical potential in phases of a stratified system (∆µ i) , (∆µ i) The equation of the spinodal corresponds to the condition

∂ 2 (∆G) ∂ϕ

2 i

=

∂(∆µ i ) ∂ϕ i

=0

[4.2.12]

where: ∆G ϕi

the Gibbs free mixing energy volume fraction of a component of a solvent.

At a critical point, binodal and spinodal coincide ∂ 2 (∆G) ∂ϕ 2i

=

∂ 3 (∆G)

[4.2.13]

∂ϕ 3i

In the elementary case of a two-component system, the Flory-Huggins theory gives the following solution:3   r ∆µ i = RT ln ϕ i +  1 − i  r  j 

   (1 − ϕ i ) + ri χ 1 ϕ i2    

[4.2.14]

where: ri, rj

numbers of segments of corresponding component.

The last equation can be solved if one takes into account the equality of chemical potentials of a component in two co-existing phases of a stratified system.  x ln ϕ′i +  1 − i  x j 

   ϕ′j + x i χ ij (ϕ′i ) 2 = ln ϕ′′i +  1 − x i     xj

  ϕ′′j + x i χ ij (ϕ′′i ) 2  

REFERENCES 1 2 3 4

P.J. Flory, J. Chem. Phys., 9, 660, (1941). P.J. Flory, J. Chem. Phys., 10, 51 (1942). P.J. Flory, Principles of polymer chemistry, Cornell University Press, Ithaca, 1953. M.L. Huggins, J. Chem. Phys., 9, 440 (1941).

[4.2.15]

132

5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

Abraham Nitzan J.H. Hildebrand and R.L. Scott, Solubility of none-electrolytes. 3rd ed., Reinhold, New-York, 1950. R.M.Masegosa, M.G. Prolonga, A. Horta, Macromolecules, 19, 1478 (1986) G.M. Bristow, J. Polym. Sci., 36, 526 (1959). J. Rehner, J. Polym. Sci., 46, 550 (1960). A.E. Nesterov, Handbook on physical chemistry of polymers. V. 1. Properties of solutions, Naukova Dumka, Kiev, 1984. I. Prigogine, The molecular theory of solutions, New York, Interscience, 1959. D. Patterson, G. Delmas, T. Somsynsky, J. Polym. Sci., 57, 79 (1962). D. Patterson, Macromolecules, 2, 672 (1969). P.J. Flory, R.A. Orwoll, A Vrij, J. Amer. Chem. Sci., 86, 3507 (1968). P.J. Flory, J. Amer. Chem. Sci., 87, 1833 (1965). A. Abe, P.J Flory, J. Amer. Chem. Sci., 87, 1838 (1965). V. Crescenzi, G. Manzini, J. Polym. Sci., C, 54, 315 (1976). A. Bondi, Physical properties of molecular crystals, liquids and glasses. New York, Wiley, 1968 M.L. Huggins, J. Phys. Chem., 74, 371 (1970). M.L. Huggins, Polymer, 12, 389 (1971). M.L. Huggins, J. Phys. Chem., 75, 1255 (1971). M.L. Huggins, Macromolecules, 4, 274 (1971). K. Thinius, Chem. Techn., 6, 330 (1954). P.J. Flory, J. Rehner, J. Chem. Phys., 11, 512 (1943). A.E. Oberth, R.S. Bruenner, J. Polym. Sci., 8, 605 (1970). U.S. Aithal, T.M. Aminabhavi, Polymer, 31, 1757 (1990). J. Bastide, S. Candau, L. Leibner, Macromolecules, 14, 319 (1981). A.E.Nesterov, Y.S. Lipatov, Phase condition of polymer solutions and mixtures, Naukova dumka, Kiev, 1987. J. Biros, L. Zeman, D.Patterson, Macromolecules, 4, 30 (1971). J.W. Kennedy , M. Gordon, G.A. Alvarez, J. Polym. Sci., 20, 463 (1975). R. Konningsveld, Brit. Polym. J., 7, 435 (1975). N.N. Avdeev, A.E. Chalych, Y.N. Moysa, R.C. Barstein, Vysokomol. soed., A22, 945 (1980).

4.3 POLAR SOLVATION DYNAMICS: THEORY AND SIMULATIONS Abraham Nitzan School of Chemistry, The Sackler Faculty of Sciences, Tel Aviv University, Tel Aviv, Israel

4.3.1 INTRODUCTION When an ion or, more generally, a charge distribution associated with a given solute is placed in a dielectric solvent, the solvent responds to accommodate this solute and to minimize the free energy of the overall system. Equilibrium aspects of this phenomenon are related to macroscopic observables such as solvation free energy and microscopic properties such as the structure of the solvation ‘shell’. Dynamical aspects of this process are manifested in the time evolution of this solvent response.1 A direct way to observe this dynamics is via the time evolution of the spectral line-shifts following a pulse excitation of a solute molecule into an electronic state with a different charge distribution.1 Indirectly, this dynamics can have a substantial effect on the course and the rate of chemical reactions that involve a redistribution of solute charges in the polar solvent environment.2 Following a brief introduction to this subject within the framework of linear response and continuum dielectric theories, this chapter describes numerical simulation studies of this process, and con-

4.3 Polar solvation dynamics

133

trasts the results obtained from such simulations with those obtained from linear response continuum models. In particular we focus on the following issues: • How well can the solvation process be described by linear response theory? • To what extent can the dynamics of the solvation process be described by continuum dielectric theory? • What are the signatures of the solute and solvent structures in the deviation of the observed dynamics from that predicted by continuum dielectric theory? • What are the relative roles played by different degrees of freedom of the solvent motion, in particular, rotation and translation, in the solvation process? • How do inertial (as opposed to diffusive) solvent motions manifest themselves in the solvation process? This chapter is not an exhaustive review of theoretical treatments of solvation dynamics. Rather, it provides, within a simple model, an exposition of the numerical approach to this problem. It should be mentioned that a substantial effort has been recently directed towards developing a theoretical understanding of this phenomenon. The starting point for such analytical efforts is linear response theory. Different approaches include the dynamical mean spherical approximation (MSA),3,4 generalized transport equations,5-8 and ad hoc models for the frequency and wavevector dependence of the dielectric response function ε(k, ω).9 These linear response theories are very valuable in providing fundamental understanding. However, they cannot explore the limits of validity of the underlying linear response models. Numerical simulations can probe non-linear effects, but are very useful also for the direct visualization and examination of the interplay between solvent and solute properties and the different relaxation times associated with the solvation process. A substantial number of such simulations have been carried out in recent years.10,11 The present account describes the methodology of this approach and the information it yields. 4.3.2 CONTINUUM DIELECTRIC THEORY OF SOLVATION DYNAMICS The Born theory of solvation applies continuum dielectric theory to the calculation of the solvation energy of an ion of charge q and radius a in a solvent characterized by a static dielectric constant, εs. The well known result for the solvation free energy, i.e., the reversible work needed to transfer an ion from the interior of a dielectric solvent to vacuum, is W =

 q2  1  − 1 2a  εs 

[4.3.1]

Eq. [4.3.1] corresponds only to the electrostatic contribution to the solvation energy. In experiments where the charge distribution on a solute molecule is suddenly changed (e.g. during photoionization of the solute) this is the most important contribution because short range solute-solvent interactions (i.e., solute size) are essentially unchanged in such processes. The origin of W is the induced polarization in the solvent under the solute electrostatic field. The time evolution of this polarization can be computed from the dynamic dielectric properties of the solvent expressed by the dielectric response function ε( ω).12 Within the usual linear response assumption, the electrostatic displacement and field are related to each other by

134

Abraham Nitzan t

D(t ) = ∫ dt ′ε(t − t ′)E (t ′)

[4.3.2]

−∞

∞

and their Fourier transforms (e.g. E ( ω) = ∫ dte − iωt E ( t )) satisfy D( ω) = ε( ω)E ( ω)

−∞

[4.3.3]

where ∞

ε(ω) ≡ ∫ dte − iωt ε(t)

[4.3.4]

0

rewriting ε(t) in the form ε(t) = 2ε e δ(t) + ~ ε(t)

[4.3.5]

we get t

D(t) = ε e E (t) + ∫ dt ′~ ε(t − t ′)E (t ′)

[4.3.6]

D(ω) = ε e E (ω) + ~ ε(ω)E (ω)

[4.3.7]

−∞

In Eq. [4.3.5] εe is the “instantaneous” part of the solvent response, associated with its electronic polarizability. For simplicity we limit ourselves to the Debye model for dielectric relaxation in which the kernel ~ ε in [4.3.5] takes the form ε − ε e − t / τD ~ ε(t) = s e τD

[4.3.8]

This function is characterized by three parameters: the electronic εe and static εs response constants, and the Debye relaxation time, τD. In this case ∞

ε(ω) = ε e + ∫ dt 0

εs − ε e − t / τ − iωt ε − εe e e = εe + s τD 1 + iωτ D

[4.3.9]

In this model a step function change in the electrostatic field E (t) = 0, t < 0; E (t) = E, t ≥ 0

[4.3.10]

leads to t

D(t ) = εeE(t ) + ∫ 0

[(

εs − εe − ( t − t ′)/ τD e E(t ′)dt ′ = εs 1− e − t / τD τD

) + ε e ]E e

− t / τD

[4.3.11]


135

For t→ 0, D(t) becomes εeE, and for t→ ∞ it is D = εsE. The relaxation process which carries the initial response to its final value is exponential, with the characteristic relaxation time, τD. The result [4.3.11] is relevant for an experiment in which a potential difference is suddenly switched on and held constant between two electrodes separated by a dielectric spacer. This means that the electrostatic field is held constant as the solvent polarization relaxes. For this to happen the surface charge density on the electrodes, i.e. the dielectric displacement D, has to change under the voltage source so as to keep the field constant. The solvation dynamics experiment of interest here is different: Here at time t = 0 the charge distribution ρ(r) is switched on and is kept constant as the solvent relaxes. In other words, the dielectric displacement D, the solution of the Poisson equation ∇D = 4πρ that corresponds to the given ρ is kept constant while the solvent polarization and the electrostatic field relax to equilibrium. To see how the relaxation proceeds in this case we start again from t

D(t) = ε e E (t) + ∫ dt ′~ ε (t − t ′) E (t ′)

[4.3.12]

−∞

take the time derivative of both sides with respect to t t

ε dD dE  d~ = εe + E (t)~ ε(0) + ∫ dt ′  E (t ′) dt dt  dt  t − t ′ −∞

[4.3.13]

use the relations ~ ε(0) = (ε s − ε e ) / τD and t

ε 1  d~ ∫−∞dt ′  dt  E (t ′) = − τ D t−t′

t

1 ∫ ~ε (t − t ′) E (t ′) = − τ (D(t) − ε E (t)) e

−∞

D

(cf Eq. [4.3.8]), to get 1 d (D − ε e E ) = − (D − εs E ) τD dt

[4.3.14]

When D evolves under a constant E, Eq. [4.3.14] implies that (d/dt)D = (-1/τD)D + constant, so that D relaxes exponentially with the time constant τD, as before. However if E relaxes under a constant D, the time evolution of E is given by ε d E =− s εe τD dt

 1   E − D  εs  

[4.3.15]

i.e. E (t) =

1 D + Ae − t / τ L εs

[4.3.16]

136

Abraham Nitzan

where A is an integration constant and τL is the longitudinal Debye relaxation time τL =

εe τD εs

[4.3.17]

The integration constant A is determined from the initial conditions: Immediately following the switch-on of the charge distribution, i.e. of D, E is given by E(t=0) = D/εe, so A = ε -1e − ε −1 D. Thus, finally, s

(

)

E (t) =

 1 1 1 D +  −  De − t / τ L εs ε ε  e s 

[4.3.18]

We see that in this case the relaxation is characterized by the time τL which can be very different from τD. For example, in water ε e / ε s ≅ 1/ 40, and while τD ≅ 10ps, τL is of the order of 0.25ps. 4.3.3 LINEAR RESPONSE THEORY OF SOLVATION DYNAMICS The continuum dielectric theory of solvation dynamics is a linear response theory, as expressed by the linear relation between the perturbation D and the response of E, Eq. [4.3.2]. Linear response theory of solvation dynamics may be cast in a general form that does not depend on the model used for the dielectric environment and can therefore be applied also in molecular theories.13,14 Let H = H0 + H′

[4.3.19]

where H0 describes the unperturbed system that is characterized by a given potential surface on which the nuclei move, and where H′ = ∑ X j F j (t)

[4.3.20]

j

is some perturbation written as a sum of products of system variables Xj and external time dependent perturbations Fj(t). The nature of X and F depend on the particular experiment: If for example the perturbation is caused by a point charge q(t) at position rj, q(t)δ(r-rj), we may identify F(t) with this charge and the corresponding Xj is the electrostatic potential operator at the charge position. For a continuous distribution ρ(r,t) of such charge we may write H ′ = ∫ d 3 rΦ (r) ρ (r , t), and for ρ(r, t) = ∑ j q j (t) δ ( r − r j ) this becomes ∑ j Φ ( r j ) q j ( t ). Alternatively we may find it convenient to express the charge distribution in terms of point moments (dipoles, quadrupoles, etc.) coupled to the corresponding local potential gradient tensors, e.g. H′ will contain terms of the form µ∇Φ and Q:∇∇Φ where µ and Q are point dipoles and quadrupoles respectively. In linear response theory the corresponding solvation energies are proportional to the corresponding products q, µ and Q: where denotes the usual observable average. For example, the average potential is proportional in linear response to q the perturbation source q. The energy needed to create the charge q is therefore ∫ dq ′ 0 ≈ (1/ 2) q 2 ≈ (1/ 2) q. Going back to the general expressions [4.3.19] and [4.3.20], linear response theory relates non-equilibrium relaxation close to equilibrium to the dynamics of equilibrium fluctu-


137

ations: The first fluctuation dissipation theorem states that following a step function change in F: F j (t) = 0, t < 0; F j (t) = F j , t ≥ 0

[4.3.21]

the corresponding averaged system’s observable relaxes to its final equilibrium value as t → ∞ according to X j (t ) − X j (∞) =

1 k BT

∑F ( X l

l

j

X l (t) − X j X l

)

[4.3.22]

where all averages are calculated with the equilibrium ensemble of H0. Applying Eq. [4.3.22] to the case where a sudden switch of a point charge q → q + ∆q takes place, we have Φ(t) − Φ(∞) =

∆q k BT

( ΦΦ(t) − Φ ) = k∆qT 2

δΦδΦ(t)

[4.3.23]

B

The left hand side of [4.3.23], normalized to 1 at t = 0, is a linear approximation to the solvation function S(t) =

Φ(t ) − Φ( ∞ ) E solv (t ) − E solv ( ∞ ) LR E solv ( 0) − E solv ( ∞ ) Φ( 0) − Φ( ∞ )

[4.3.24]

and Eq. [4.3.23] shows that in linear response theory this non equilibrium relaxation function is identical to the equilibrium correlation function S(t) LR C(t) ≡

δΦ( 0) δΦ(t ) δΦ 2

[4.3.25]

C(t) is the time correlation function of equilibrium fluctuations of the solvent response potential at the position of the solute ion. The electrostatic potential in C(t) will be replaced by the electric field or by higher gradients of the electrostatic potential when solvation of higher moments of the charge distribution is considered. The time dependent solvation function S(t) is a directly observed quantity as well as a convenient tool for numerical simulation studies. The corresponding linear response approximation C(t) is also easily computed from numerical simulations, and can also be studied using suitable theoretical models. Computer simulations are very valuable both in exploring the validity of such theoretical calculations, as well as the validity of linear response theory itself (by comparing S(t) to C(t)). Furthermore they can be used for direct visualization of the solute and solvent motions that dominate the solvation process. Many such simulations were published in the past decade, using different models for solvents such as water, alcohols and acetonitrile. Two remarkable outcomes of these studies are first, the close qualitative similarity between the time evolution of solvation in different simple solvents, and second, the marked deviation from the simple exponential relaxation predicted by the Debye relaxation model (cf. Eq. [4.3.18]). At least two distinct relaxation modes are

138

Abraham Nitzan

observed, a fast Gaussian-like component and a slower relaxation mode of an exponential character which may correspond to the expected Debye relaxation. In what follows we describe these and other features observed in computer simulations of solvation dynamics using simple generic model dielectric solvents. 4.3.4 NUMERICAL SIMULATIONS OF SOLVATION IN SIMPLE POLAR SOLVENTS: THE SIMULATION MODEL11a The simplest simulated system is a Stockmayer fluid: structureless particles characterized by dipole-dipole and Lennard-Jones interactions, moving in a box (size L) with periodic boundary conditions. The results described below were obtained using 400 such particles and in addition a solute atom A which can become an ion of charge q embedded in this solvent. The long range nature of the electrostatic interactions is handled within the effective dielectric environment scheme.15 In this approach the simulated system is taken to be surrounded by a continuum dielectric environment whose dielectric constant ε′ is to be chosen self consistently with that computed from the simulation. Accordingly, the electrostatic potential between any two particles is supplemented by the image interaction associated with a spherical dielectric boundary of radius Rc (taken equal to L/2) placed so that one of these particles is at its center. The Lagrangian of the system is given by

(

)

L R, R& , µ , µ& = −

N 1 1 N 1 N I 1 N MAR& A2 + M ∑ R& i2 + M ∑ 2 µ& 2i − ∑VijIJ (Rij ) − ∑ViAIJ (RiA) − 2 2 i =1 2 i =1 µ 2 i≠j i =1

(

N N 1 N DD V (Ri , Rj , µ i , µ j ) − ∑V AD (RA, Ri , µ i ) − ∑ λ i µ 2i − µ 2 ∑ 2 i≠j i =1 i =1

)

[4.3.26]

where N is the number of solvent molecules of mass M, µ dipole moment, and I moment of inertia. RA and Ri are positions of the impurity atom (that becomes an ion with charge q) and a solvent molecule, respectively, and Rij is |Ri - Rj|. VLJ, VDD and VAD are, respectively, Lennard-Jones, dipole-dipole, and charge-dipole potentials, given by VijLJ (R) = eε D

[(σ

D

/ R)

12

− (σ D / R)

6

]

[4.3.27]

(VijLJ is of the same form with σA and εA replacing σD and εD) and

(

)

V DD Ri , R j , µ i , µ j =

(

µ i µ j − 3(nµ i ) nµ j R

3 ij

) − 2(ε′ − 1)

(2ε′ + 1) RC3

µ iµ j

[4.3.28]

where n = (Ri - Rj)/Rij,  1 2(ε′ − 1) V AD (Ri , µ i , RA ) = q  3 − 3 R  ij (2ε′ + 1)RC

  µ (R − R ) A  i i 

[4.3.29]

The terms containing ε′ in the electrostatic potentials VDD and VAD are the reaction field image terms.15,16 The last term in Eq. [4.3.26] is included in the Lagrangian as a constraint, in order to preserve the magnitude of the dipole moments ( µ i = µ ) with a SHAKE


139

like algorithm.17 In this representation the mass, M, and the moment of inertia, I, of the solvent molecules are independent parameters, which makes it possible to study the relative importance of translational and rotational motions in the solvation process without affecting other potentially relevant parameters such as the molecular size. The time evolution is done using the velocity Verlet algorithm, with the value of λ(t) determined as in the SHAKE algorithm, and with the Andersen18 thermalization used to keep the system at constant temperature. For the Stockmayer solvent, the initial molecular parameters are taken to approximate the CH3Cl molecule: σD = 4.2 Å, εD = 195K, M = 50 amu, I = 33.54 amu Å2, and µ = 1.87 D. The parameters taken for the solvated ion are MA = 25 amu, σA = 3.675 Å, and εA = 120K, q is taken to be one electron charge e. These parameters can be changed so as to examine their effect on the solvation dynamics. Most of the results described below are from simulations done at 240K, and using L = 33.2Å for the edge length of the cubic simulation cell was, corresponding to the density ρ = 1.09×10-2 Å-3, which is the density of CH3Cl at this temperature. In reduced units we have for this choice of parameters ρ * ≡ ρσD3 = 0.81, 3 1/ 2 * µ ≡ µ εD σD = 132, . T* ≡ kBT/εD =1.23, and I* ≡ I(Mσ2) = 0.038. A simple switching function

(

)

f(R) =

8

1 R < Rs (Rc − R ) / (Rc − Rs ) Rs < R < Rc 0 R > Rc

[4.3.31]

is used to smoothly cut off this electrostatic potential. Rc and Rs are taken to be Rc = L/2 and Rs = 0.95Rc. Under these simulation conditions the pressure fluctuates in the range 500±100 At.The dielectric constant is computed from pure solvent simulations, using the expression33

(ε − 1)(2ε′ + 1) 2ε′ + ε

=

1 PP(Rc ) k BTRc3

[4.3.32]

where N

P = ∑µ i

[4.3.33]

i =1

and P(Rc ) =

1 N N ∑ ∑'µ k N j =1 k =1

[4.3.34]

where the prime on the inner summation indicates the restriction Rjk < Rc. The result for our solvent is ε = 17, compared with ε(CH3Cl) = 12.6 at 253K. After evaluating ε in this way the external dielectric constant ε′ is set to ε and the computation is, in principle, repeated until convergence, i.e., until the evaluated ε is equal to the environmental ε′. In fact, we have found that our dynamical results are not sensitive to the magnitude of ε′.

140

Abraham Nitzan

Figure 4.3.1. The electrostatic response potential induced by the solvent at the position of the solute ion, as a function of the solute charge. Dashed line - the Stockmayer-CH3Cl model described in Section 4. Full and dotted lines, model polyether solvents described in the text. [From Ref. 11b].

Figure 4.3.2. The linear response relaxation function C(t) (dashed and dotted lines] and the non-equilibrium solvation function S(t) (solid line) computed for the Stockmayer-CH3Cl model described in Section 4. In the nonequilibrium simulation the ion charge is switched on at t = 0. The dotted and dashed lines represent C(t) obtained from equilibrium simulations with uncharged and charged ion, respectively. [From Ref. 11a].

A typical timestep for these simulation is 3fs. In the absence of thermalization this gives energy conservation to within 10-4 over ~80,000 time steps. After equilibrating the system at 240K, the equilibrium correlation function C(t) is evaluated from equilibrium trajectories with both a charged (q = e) and an uncharged (q = 0) impurity atom. The non-equilibrium solvation function S(t) can also be computed from trajectories that follow a step function change in the ion charge from q = 0 to q = e. These calculations are done for the CH3Cl solvent model characterized by the above parameters and for similar models with different parameters. In particular, results are shown below for systems characterized by different values of the parameter15 p ′ = I / 2Mσ 2

[4.3.35]

which measures the relative importance of rotational and translational solvent motions. 4.3.5 NUMERICAL SIMULATIONS OF SOLVATION IN SIMPLE POLAR SOLVENTS: RESULTS AND DISCUSSION The dashed line of Figure 4.3.1 shows the equilibrium solvent induced electrostatic potential Φ at the position of the ion, as a function of the ion charge q obtained for the Stockmayer-CH3Cl model described in Section 4. Clearly the solvent response is linear with q all the way up to q = e, with slight deviations from linearity starting at q > e. The slope (~4) of the linear dependence of the dashed line in Figure 1 (for q < e) is considerably smaller from that obtained from Φ = q / aε s (taking a = σA/2 gives a slope of 7.4) that is used to get Eq. [4.3.1]. A more advanced theory of solvation based on the mean spherical approximation predicts (using σA and σD for the diameters of the ion and the solvent, respectively) a slope of 4.6. The linearity of the response depends on the nature of the solvent. As examples Figure 4.3.1 also shows results obtained for models of more complex solvents, H(CH2OCH2)nCH3


141

with n = 1 (ethyl methyl ether, full line) and n = 2 (1,2-methoxy ethoxy ethane, dotted line). In these solvents the main contribution to the solvation energy of a positive ion comes from its interaction with solvent oxygen atoms. Because of geometric restriction the number of such atoms in the ion’s first solvation shell is limited, leading to a relatively early onset of dielectric saturation. Figure 4.3.2 shows the time evolution of the solvation functions C(t), Eq. [4.3.25] Figure 4.3.3. The function S(t) obtained with ε ′ = 17.0 and S(t) (Eq. [4.3.24]). C(t) is evaluated (solid line, same as in Figure 4.3.2), together with the from an equilibrium trajectory of 220 ps for same function obtained with ε ′ = 1.0 (dashed line). ε ′ is a system consisting of the solvent and a the continuum dielectric constant associated with the recharged or uncharged atom. The action field boundary conditions. [From Ref. 11a]. nonequilibrium results for S(t) are averages over 25 different trajectories, each starting from an initial configuration taken from an equilibrium run of an all-neutral system following switching, at t = 0, of the charge on the impurity atom from q = 0 to q = e. These results show a large degree of similarity between the linear response (equilibrium) and nonequilibrium results. Both consist of an initial fast relaxation mode that, at closer inspection is found to be represented well by a Gaussian, exp[-(t/τ)2], followed by a relatively slow residual relaxation. The initial fast part is more pronounced in C(t). The latter is also characterized by stronger oscillations in the residual part of the relaxation. The fact that the linear response results obtained for equilibrium simulations with an uncharged solute and with a solute of charge q are very similar give further evidence to the approximate validity of linear response theory for this systems. The sensitivity of these results to the choice of boundary conditions is examined in Figure 4.3.3. We note that the use of reaction field boundary conditions as implemented here is strictly valid only for equilibrium simulations, since the dynamic response of the dielectric continuum at R > Rc is not taken into account. One could argue that for the short-time phenomena considered here, ε′ should have been taken smaller than the static dielectric constant of the system. Figure 4.3.3 shows that on the relevant time scale our dynamical results do not change if we take ε′ = 1 instead of ε′ = ε = 17. (The absolute solvation energy does depend on ε′, and replacing ε′ = 17 by ε′ = 1 changes it by ≅ 5%.) In the simulations described so far the solvent parameters are given by the aforementioned data. For these, the dimensionless parameter p′, Eq. [4.3.35], is 0.019. In order to separate between the effects of the solvent translational and rotational degrees of freedom, we can study systems characterized by other p′ values. Figure 4.3.4 shows results obtained for p′ = 0 (dotted line), 0.019 (solid line), 0.25 (dashed line), and ∞ (dashed-dotted line). Except for p′ = 0, these values were obtained by changing the moment of inertia I, keeping M=50 amu. The value p′ = 0 was achieved by taking M = MA = ∞ and I=33.54 amu Å2. Note that the values p′ = 0 and p′ = ∞ correspond to models with frozen translations and frozen rotations, respectively. Figures 4.3.4(a) and 4.3.4(b) show, respectively, the solvation energy Esolv(t) and the solvation function S(t) obtained for these different systems. The following points are noteworthy:

142

Abraham Nitzan Table 4.3.1. Relaxation times τ obtained from fitting the short time component of the solvation function to the function S(t) = exp[-(t / τ )2 ]. The fitting is done for S(t) > 0.3 [From Ref. 11a]. p′

M amu

I amu Å2

τ ps

0.0

∞

33.54

0.206

0.019

50

33.54

0.170

0.125

50

220.5

0.347

0.250

50

441.0

0.421

(1) The asymptotic (t→ ∞) values of Esolv (Figure 4.3.4a) are different for p′ = 0 (M = ∞) and p′ = ∞ (I = ∞) then in the other cases because of the freezing of solvent translations and rotations, respectively. Figure 4.3.4. The solvation energy, Esolv (a) and the Note, however, that the I = ∞ curves connon-equilibrium solvation function S(t) (b), plotted verge very slowly (the solvent compensates against time (after switching the ion charge from 0 to e at for the lack of rotations by bringing into the t = 0) for different solvent models characterized by the neighborhood of the solute solvent moleparameter p ′ (Eq. [4.3.35]). Dotted line, p ′ = 0; solid line, p ′=0.019; dashed line, p ′=0.25; dashed-dotted line, p ′=8. cules with the “correct” orientation) and [From Ref. 11a]. probably did not reach its asymptotic value in Figure 3.4.4a. (2) Except for the rotationless system (p′ = ∞) all the other systems exhibit a bimodal relaxation, with a fast relaxation component that accounts for most of the solvation energy. The relaxation of the rotationless solvent is exponential (a fit to exp(-t/τ) yields τ = 2.2 ps). (3) A closer look at the fast component in the finite p′ systems shows a Gaussian behavior, a fit to exp[-(t/τ)2] yields the τ values summarized in Table 4.3.1. τ increases with increasing solvent moment of inertia (recall that this is how p′ is changed for p′ > 0, still for the range of p′ studied, it stays distinct from the long component. (4) The oscillations and the thermal noise seen in the relatively small slow relaxation component make it difficult to estimate the long relaxation time. A fit of the long time component for the p′ = 0.019 case to an exponential relaxation exp(-t/τ) yields τ ≅ 2.7 ± 0.7 ps. The long-time components in the other systems relax on similar time scales. The nature of this fast relaxation component of the solvation dynamics has been discussed extensively in the past decade.19 Carter and Hynes20 were the first to suggest that this initial relaxation is associated with inertial, as opposed to diffusive, solvent motion. In this mode of motion solvent molecules move under the suddenly created force-field without their mutual interactions and consequent energy exchange having a substantial influence on this fast time scale. Simulations and analytical calculations19,21,22 have confirmed this assertion for simple polar fluids.


143

Figure 4.3.6. The time dependence of the average angle between the molecular dipoles and between the corresponding radius vectors to the ion center, associated with molecules in the three different solvation shells defined in the text, plotted against time following the switching on of the ionic charge. [From Ref. 11a].

Next we examine the relative contributions of different solvation shells to the solvation process. This issue is important for elucidating the solvation mechanism, Figure 4.3.5.(a) The solvation energy E(t) and (b) the and has been under discussion since an solvation function S(t) associated with the three solva- early remark made by Onsager23 that the tion shells defined in the text, plotted against time after the ion charge is switched on, for the system with shorter time scales are associated mostly p ′=0.019. Solid line, nearest shell; dotted line, second with solvent layers further away from the shell, dashed line, outer shell. [From Ref. 11a]. solute, and that the longer ~ τD times are associated with the individual response of solvent molecules nearest to the solute. From the structure of the solute-solvent radial distribution function of the simulated system one can estimate11a that the first solvation shell about the solute consists of the eight nearest neighbor solvent particles at distance closer than 5.5 Å from the solute center, and the second solvation shell encompasses the next nearest 26 solvent particles at distance smaller than ~10 Å from the solute center. Taking the rest of the solvent particles in the simulation box as the “third solvation shell”, Figure 4.3.5 shows the contributions of these layers to the time evolution of the solvation energy and of the solvation function. It seems that the fast component in these time evolutions is faster for the contribution from the first solvation shell. The same shell also shows a distinct slow component which is much smaller or absent in the contribution from the further shells. Also note that the solvation energy is dominated by the first solvation shell: the first, second, and third shells contribute ~67%, 24%, and 9%, respectively, to the solvation energy. The fast relaxation component accounts for ~80% of the solvation energy. It should be kept in mind, however, that the contribution from outer shell molecules is suppressed by the finite size of the simulated system. Finally, the nature of the motion that gives rise to the fast relaxation component is seen in Figure 6, which depicts as functions of time the average angles between the molecular dipoles in the different solvation shells and between the corresponding radius vectors to the

144

Abraham Nitzan

ion centers. These results are for the p′ = 0.019 system; the other systems with p′ < ∞ show qualitatively similar behavior. Generally, the time evolution of the angular motion is similar to that of the solvation energy. Typical to the present system that represents simple polar solvents, the fast relaxation component is associated with the initial relaxation of the orientational structure in the solvation layers close to the solute. 4.3.6 SOLVATION IN COMPLEX SOLVENTS The previous sections have focused on a generic model of a very simple solvent, in which solvation dynamics is determined by molecular translations and reorientations only. These in turn are controlled by the solvent molecular mass, moment of inertia, dipole moment and short-range repulsive interactions. When the solvent is more complex we may expect specific structures and interactions to play significant roles. Still, numerical simulations of solvation dynamics in more complex systems lead to some general observations: (a) In large molecular solvents, solvation may be associated with binding of the solute to particular solvents sites. As seen in Figure 4.3.1, deviations from linearity in the solvent response potential are associated with the fact that the fraction of polar binding sites constitutes a relatively small fraction of the solvent molecule. This deviation from linearity shows itself also in the solvation dynamics. Figure 4.3.7 shows the linear response functions and the non-equilibrium solvation function, C(t) and S(t), respectively, computed as before, for the di-ether H(CH2OCH2)2CH3 solvent. Details of this simulations are given in Ref.11b. If linear response was a valid approximation all the lines in Figure 4.3.7: The two lines for C(t) that correspond to q=0 and q=1, and the two lines for S(t) Figure 4.3.7. The solvation and response functions, S(t) for the processes q=0→q=1 and the process and C(t), respectively, for solvation of a spherical ion in a model for the solvent 1,2-methoxy ethoxy ethane, q=1→q=0, would coalesce. The marked H(CH2OCH2)2CH3. Full line: S0 →1(t); dotted line: S1→ 0 (t); differences between these lines shows that dashed line: C(t)|q=0 and dotted-dashed line: C(t)|q=1. linear response theory fails for this system. [From Ref. 11b]. (b) Linear response theory was also shown to fail for low-density solvents (e.g. near and above the liquid-gas critical point11c,24). In this case the origin of the non-linearity is the large rearrangement in the solvent structure near the solute during the solvation process. This rearrangement is associated with a local density change in such highly compressible low-density solvents. (c) Similarly, solvation in mixed solvents usually involve large rearrangement of solvent structure near the solute because the latter usually have a higher affinity for one of the solvent components. Solvation in electrolyte solutions provides a special example.25,26 In this case the solvent dynamics about the newly created charge distribution is not much different than in the pure dielectric solution, however in addition the mobile ions rearrange about this charge distribution, and on the timescale of this process linear response theory fails.27 (d) In the situations discussed in (b) and (c) above, new dynamical processes exist: While the dielectric response in normal simple polar solvents is dominated by molecular ro-


145

tations, the motions that change the local structure about the solute are usually dominated by solvent translation. This gives rise to a new, usually slower, relaxation components. Solvation in electrolyte solutions clearly shows this effect: In addition to the dielectric response on the picosecond timescale, a much slower relaxation component is observed on the nanosecond scale.25 Numerical simulations have identified the origin of this relaxation component as the exchange between a water molecule and an ion in the first solvation shell about the solute.26 Finally, it is intuitively clear that in large molecule complex solvents simple molecular rotation as seen in Figure 4.3.6 can not be the principal mode of solvation. Numerical simulations with polyether solvents show that instead, hindered intramolecular rotations that distort the molecular structure so as to bring more solvating sites into contact with the ion dominate the solvation dynamics.11b The bi-modal, and in fact multi-modal, character of the solvation is maintained also in such solvents, but it appears that the short time component of this solvation process is no longer inertial as in the simple small molecule solvents.11 4.3.7 CONCLUSIONS Numerical simulations of solvation dynamics in polar molecular solvents have been carried out on many models of molecular systems during the last decade. The study described in sections 4.3.4-4.3.5 focused on a generic model for a simple polar solvent, a structureless Stockmayer fluid. It is found that solvation dynamics in this model solvent is qualitatively similar to that observed in more realistic models of more structured simple solvents, including solvents like water whose energetics is strongly influenced by the H-bond network. In particular, the bimodal nature of the dynamics and the existing of a prominent fast Gaussian relaxation component are common to all models studied. Such numerical simulations have played an important role in the development of our understanding of solvation dynamics. For example, they have provided the first indication that simple dielectric continuum models based on Debye and Debey-like dielectric relaxation theories are inadequate on the fast timescales that are experimentally accessible today. It is important to keep in mind that this failure of simple theories is not a failure of linear response theory. Once revised to describe reliably response on short time and length scales, e.g. by using the full k and ω dependent dielectric response function ε(k,ω), and sufficiently taking into account the solvent structure about the solute, linear response theory accounts for most observations of solvation dynamics in simple polar solvents. Numerical simulations have also been instrumental in elucidating the differences between simple and complex solvents in the way they dynamically respond to a newly created charge distribution. The importance of translational motions that change the composition or structure near the solute, the consequent early failure of linear response theory in such systems, and the possible involvement of solvent intramolecular motions in the solvation process were discovered in this way. We conclude by pointing out that this report has focused on solvation in polar systems where the solvent molecule has a permanent dipole moment. Recently theoretical and experimental work has started on the dynamics of non-polar solvation.28 This constitutes another issue in our ongoing effort to understand the dynamics of solvation processes. REFERENCES 1

For recent reviews see M. Maroncelli, J. Mol. Liquids, 57, 1 (1993); G.R. Fleming and M. Cho, Ann. Rev. Phys. Chem., 47, 109 (1996).

146

2

3 4 5 6 7 8 9 10

11

12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

Christian Wohlfarth See, e.g., J.T. Hynes, in Ultrafast Dynamics of Chemical Systems, edited by J.D. Simon, Kluwer, Dordrecht, 1994, pp. 345-381; L.D. Zusman, Zeit. Phys. Chem., 186, 1 (1994); S. Roy and B. Bagchi, J. Chem. Phys., 102, 6719; 7937 (1995). P. G. Wolynes, J. Chem. Phys., 86, 5133 (1987). I. Rips, J. Klafter and J. Jortner, J. Chem. Phys., 88, 3246 (1988); 89, 4288 (1988). D. F. Calef and P. G. Wolynes, J. Chem. Phys., 78, 4145 (1983). B. Bagchi and A. Chandra, J. Chem. Phys., 90, 7338 (1989); 91, 2594 (1989); 97, 5126 (1992). F.O Raineri, H. Resat, B-C Perng, F. Hirata and H.L. Friedman, J. Chem. Phys., 100, 1477 (1994) R. F. Loring and S. Mukamel, J. Chem. Phys., 87, 1272 (1987); L. E. Fried and S. Mukamel, J. Chem. Phys., 93, 932 (1990). A. A. Kornyshev. A. M. Kuznetsov, D. K. Phelps and M. J. Weaver, J. Chem. Phys., 91, 7159 (1989). See, e.g. M. Maronelli and G.R. Fleming JCP, 89, 5044 (1988); M. Maroncelli, J. Chem. Phys., 94, 2084 (1991); Perera and Berkowitz, J. Chem. Phys., 96, 3092 (1992); P.V. Kumar and M. Maroncelli, J. Chem. Phys., 103, 3038 (1995). (a) E. Neria and A. Nitzan, J. Chem. Phys., 96, 5433 (1992). (b) R. Olender and A. Nitzan, J. Chem. Phys., 102, 7180 (1995). (c) P. Graf and A. Nitzan, Chem. Phys., 235, 297(1998). A. Mozumder in Electron-solvent and anion- solvent interactions, L. Kevan and B. Webster, Editors, Elsevier, Amsterdam, 1976. M. Maronelli and G.R. Fleming in Ref. 11. E.A. Carter and J.T. Hynes, JCP, 94, 2084 (1991). J. W. de Leeuw, J. W. Perram and E. R. Smith, Annu. Rev. Phys. Chem., 37, 245 (1986). C. J. F. Böttcher and P. Bordewijc, Theory of Electric Polarization, 2nd. ed. Elsevier, Amsterdam, 1978, Vol. 2, Chap. 10. M. P. Allen and D. J. Tildesely, Computer Simulation of Liquids, Oxford, London, 1989). H. C. Andersen, J. Chem. Phys., 72, 2384 (1980). See, e.g. M. Maroncelli, in Ref. 11. E. A. Carter and J.T. Hynes, J. Chem. Phys., 94, 5961 (1991). M. Maroncelli, V.P. Kumar and A. Papazyan, J. Phys. Chem., 97, 13 (1993). L. Perera and M.L. Berkowitz, J. Chem. Phys., 97, 5253 (1992). L. Onsager, Can. J. Chem., 55, 1819 (1977). R. Biswas and B. Bagchi, Chem. Phys. Lett., 290 (1998). V. Itah and D. Huppert, Chem. Phys. Lett., 173, 496 (1990); E. Bart and D. Huppert, ibid. 195, 37 (1992). C.F. Chapman and M. Maroncelli, J. Phys. Chem., 95, 9095 (1991). E. Neria and A. Nitzan, J. Chem. Phys., 100, 3855 (1994). B.M. Ladanyi and M. Maroncelli, J. Chem. Phys., 109, 3204 (1998); M. Berg, J. Phys. Chem., A102, 17 (1998); Chem. Phys., 233, 257, (1998).

4.4 METHODS FOR THE MEASUREMENT OF SOLVENT ACTIVITY OF POLYMER SOLUTIONS Christian Wohlfarth Martin-Luther-University Halle-Wittenberg, Institute of Physical Chemistry, Merseburg, Germany, e-mail: [email protected]

4.4.1 INTRODUCTION Knowledge of solvent activities in polymer solutions is a necessity for a large number of industrial and laboratory processes. Such data are an essential tool for understanding the thermodynamic behavior of polymer solutions, for studying their intermolecular interactions and for getting insights into their molecular nature. Besides, they are the necessary basis for any development of theoretical thermodynamic models. Scientists and engineers in academic and industrial research need such data.

4.4 Measurement of solvent activity

147

Solvent activities of polymer solutions have been measured for about 60 years now. However, the database for polymer solutions is still modest, in comparison with the enormous amount of data available for mixtures of low-molecular substances. Three explicit databases have been published in the literature up to now.1-3 The database prepared by Wen Hao et al.1 is summarized in two parts of the DECHEMA Chemistry Data Series. Danner and High2 provided a database and some calculation methods on a floppy disk with their book. Wohlfarth3 prepared the most complete data collection regarding vapor-liquid equilibrium data of polymer solutions. His annually updated electronic database is not commercially available; however, personal requests can be made via his e-mail address given above. Some implicit databases are provided within the Polymer Handbook4 by Schuld and Wolf 5 or by Orwoll6 and in two papers prepared earlier by Orwoll.7,8 These four sources list tables of Flory’s χ-function and tables where enthalpy, entropy or volume changes, respectively, are given in the literature for a large number of polymer solutions. The tables of second virial coefficients of polymers in solution, which were prepared by Lechner and coworkers9 (also provided in the Polymer Handbook), are a valuable source for estimating the solvent activity in the dilute polymer solution. Bonner reviewed vapor-liquid equilibria in concentrated polymer solutions and listed tables containing temperature and concentration ranges of a certain number of polymer solutions.10 Two CRC-handbooks prepared by Barton list a larger number of thermodynamic data of polymer solutions in form of polymer-solvent interaction or solubility parameters.11,12 An up-to-date list of all polymer-solvent systems for which solvent activities or vapor pressures from vapor-liquid equilibrium measurements were published in the literature is provided in Appendix 4.4.A of this Subchapter 4.4 (please see below). Solvent activities in polymer solutions can be determined by rather different techniques. However, no one is really a universal method but covers a certain concentration range of the polymer solution. Figure 4.4.1 explains in short the situation. Corresponding to the different regions in the diagram, different experimental techniques were used for the measurement of the solvent activity in a homogeneous polymer solution: (i) Solvent activities of highly diluted polymer solutions can be obtained from scattering methods such as light scattering, small angle X-ray scattering and small angle neutron scattering via the second osmotic virial coefficient, which is often related to investigations for polymer characterization. These methods are able to resolve the very small difference between the thermodynamic limit of 1.0 for the activity of the pure solvent and the actual value of perhaps 0.9999x at the given (very low) polymer concentration. (ii) Solvent activities of polymer solutions with polymer concentrations up to about 30 wt% can be measured by osmometry (membrane as well as vapor-pressure osmometry), light scattering, ultracentrifuge (of course, all these methods can also be applied for polymer characterization and can be extrapolated to zero polymer concentration to obtain the second virial coefficient), and differential vapor pressure techniques. Cryoscopy and ebulliometry can also be used to measure solvent activities in dilute and semidilute polymer solutions, but with limited success only. (iii) The concentrated polymer solution between 30 and 85 wt% is covered by vapor pressure measurements which were usually performed by various isopiestic sorption meth-

148

Christian Wohlfarth

Figure 4.4.1. Typical isotherm for the solvent activity of a homogeneous binary polymer solution.

ods. The ultracentrifuge can also be applied for solutions up to 80 wt% polymer, but this was only scarcely done in the literature. (iv) A special problem are polymer solutions with concentrations higher than 90 wt% up to the limit of the region of Henry’s law. For this purpose, the inverse gas-liquid chromatography (IGC) is the most useful method. Measurements can be made at infinite dilution of the solvent for determining the activity coefficient at infinite dilution or Henry’s constant, but IGC can also be performed at finite solvent concentrations up to 10-15 wt% of the solvent to get solvent activities for highly concentrated polymer solutions. Some sorption experiments in this concentration range were reported by piezoelectric quartz crystal technique; however, thermodynamic equilibrium absorption is difficult to obtain, as discussed below. At least, melting point depression can be applied in some cases for small amounts of solvents in semicrystalline polymers, but obtaining reliable results seems to be difficult. (v) There is another possibility to measure solvent activities in polymer solutions if the state of the solution is inhomogeneous, i.e., for the region of liquid-liquid equilibrium. Binodal and/or spinodal curves can be reduced to solvent activity data by means of a thermodynamic ansatz for the Gibbs free energy of mixing in dependence on temperature, concentration (and pressure if necessary), which has to be solved according to thermodynamic equilibrium conditions. In the case of polymer networks, swelling equilibria can be measured instead. The solvent activity in a swollen network arises from two parts, a mixing part with the (virtually) infinite-molar-mass polymer, and a contribution from elastic network deformation. The second follows from statistical theory of rubber elasticity and also needs certain model approximations for data reduction.


149

In summary, investigations on vapor-liquid equilibrium of polymer solutions are the most important source for obtaining solvent activities in polymer solutions. Therefore, emphasis is laid in this subchapter on the experimental methods, which use this equilibrium. Reviews on experimental methods, sometimes including tables with thermodynamic data were prepared more or less continuously during the last three decades. Especially methods and results of the application of IGC to polymers and polymer solutions are carefully reviewed.13-25 Special reviews on determining solvent activities from various scattering techniques could not be found. However, there is a large number of reviews and books on scattering methods and their applications. Some references may give a starting point for the interested reader.26-32 Experimental techniques for vapor-pressure measurements were reviewed in the paper by Bonner.10 Ebulliometry, cryoscopy and vapor-pressure osmometry were reviewed by Cooper,33 Glover,34 Mays and Hadjichristidis,40 and a recent summary can be found in a new book edited by Pethrick and Dawkins.26 Reviews that account for the measurement of thermodynamic data from sedimentation equilibria using the ultracentrifuge are given by Fujita,35 Harding et al.36 or Munk.37 An overview on membrane osmometry was given by Adams,38 Tombs and Peacock39 or Mays and Hadjichristidis,40 and a recent summary can again be found in the book edited by Pethrick and Dawkins.26 Reviews on liquid-liquid demixing of polymer solutions will not be summarized in detail here, some references should be enough for a well-based information.41-45 A short summary on equipment and thermodynamic equations of most techniques was given in Danner’s handbook.2 Finally, the classical books on polymer solutions written by Flory,46 by Huggins,47 and by Tompa48 must not be forgotten for the historical point of view on the topic of this subchapter. 4.4.2 NECESSARY THERMODYNAMIC EQUATIONS Here, the thermodynamic relations are summarized which are necessary to understand the following text. No derivations will be made. Details can be found in good textbooks, e.g., Prausnitz et al.49 The activity of a component i at a given temperature, pressure, and composition can be defined as the ratio of the fugacity of the solvent at these conditions to the solvent fugacity in the standard state; that is, a state at the same temperature as that of the mixture and at specified conditions of pressure and composition:

(

a i (T, P, x) ≡ f i (T, P, x) / f i T, P 0 , x 0

)

[4.4.1a]

where: ai T P x fi

activity of component i absolute temperature pressure mole fraction fugacity of component i

In terms of chemical potential, the activity of component i can also be defined by:

(

 µ i (T, P, x) − µ i T, P 0 , x 0 a i (T, P, x) ≡ exp RT 

)   

[4.4.1b]

150

Christian Wohlfarth

where: µi R

chemical potential of component i gas constant

P0 and x0 denote the standard state pressure and composition. For binary polymer solutions the standard state is usually the pure liquid solvent at its saturation vapor pressure at T. The standard state fugacity and the standard state chemical potential of any component i are abbreviated in the following text by their symbols f i0 and µ 0i , respectively. Phase equilibrium conditions between two multi-component phases I and II require thermal equilibrium, T I = T II

[4.4.2a]

mechanical equilibrium, P I = P II

[4.4.2b]

and the chemical potential of each component i must be equal in both phases I and II. µ Ii = µ IIi

[4.4.3]

For Equation [4.4.3] to be satisfied, the fugacities of each component i must be equal in both phases. f i I = f i II

[4.4.4]

Applying fugacity coefficients, the isochemical potential expression leads to: φIi x iI = φIIi x iII

[4.4.5]

where: φi

fugacity coefficient of component i.

Fugacity coefficients can be calculated from an equation of state by: ln φi =

1 RT

 ∂P ∫  ∂n i V   ∞

 RT  PV   dV − ln − V  RT T ,V , nj 

[4.4.6]

where a pressure explicit equation of state is required to use Equation [4.4.6]. Not all equations of state for polymers and polymer solutions also are valid for the gaseous state (see section in Subchapter 4.4.4), however, and a mixed gamma-phi approach is used by applying Equation [4.4.7]. Applying activity coefficients in the liquid phase, the isochemical potential expression leads, in the case of the vapor-liquid equilibrium (superscript V for the vapor phase and superscript L for the liquid phase), to the following relation: φVi y i P = γ i x iL f i 0

[4.4.7]

where: yi γi

mole fraction of component i in the vapor phase with partial pressure Pi = yiP activity coefficient of component i in the liquid phase with activity a i = x Li γ i


151

or in the case of liquid-liquid equilibrium to γ Ii x iI f i 0 I = γ IIi x iII f i 0 II

[4.4.8]

If the standard state in both phases is the same, the standard fugacities cancel out in Equation [4.4.8]. Equation [4.4.8] also holds for solid-liquid equilibria after choosing appropriate standard conditions for the solid state, but they are of minor interest here. All expressions given above are exact and can be applied to small molecules as well as to macromolecules. The one difficulty is having accurate experiments to measure the necessary thermodynamic data and the other is finding correct and accurate equations of state and/or activity coefficient models to calculate them. Since mole fractions are usually not the concentration variables chosen for polymer solutions, one has to specify them in each case. The following three quantities are most frequently used: mass fractions volume fractions segment (hard-core volume) fractions

∑m ϕ =n V / ∑n V ψ =n V / ∑n V

wi = m i / i

i

i

i

i

k

k

* i

k

k

* k

[4.4.9a] [4.4.9b] [4.4.9c]

where: mi ni Vi Vi*

mass of component i amount of substance (moles) of component i molar volume of component i molar hard-core (characteristic) volume of component i.

With the necessary care, all thermodynamic expressions given above can be formulated with mass or volume or segment fractions as concentration variables instead of mole fractions. This is the common practice within polymer solution thermodynamics. Applying characteristic/hard-core volumes is the usual approach within most thermodynamic models for polymer solutions. Mass fraction based activity coefficients are widely used in Equations [4.4.7 and 4.4.8] which are related to activity by: Ω i = a i / wi

[4.4.10]

where: Ωi ai wi

mass fraction based activity coefficient of component i activity of component i mass fraction of component i

Classical polymer solution thermodynamics often did not consider solvent activities or solvent activity coefficients but usually a dimensionless quantity, the so-called Flory-Huggins interaction parameter χ.44,45 χ is not only a function of temperature (and pressure), as was evident from its foundation, but it is also a function of composition and polymer molecular mass.5,7,8 As pointed out in many papers, it is more precise to call it χ-function (what is in principle a residual solvent chemical potential function). Because of its widespread use and its possible sources of mistakes and misinterpretations, the necessary relations must be included here. Starting from Equation [4.4.1b], the difference between the chemical potentials of the solvent in the mixture and in the standard state belongs to the first

152

Christian Wohlfarth

derivative of the Gibbs free energy of mixing with respect to the amount of substance of the solvent:  ∂n∆ mix G   ∆µ 1 = µ 1 − µ 10 =   ∂n1 T , p, nj≠1

[4.4.11]

where: ni n ∆ mixG

amount of substance (moles) of component i total amount of substance (moles) of the mixture: n = Σni molar Gibbs free energy of mixing.

For a truly binary polymer solutions, the classical Flory-Huggins theory leads to:46,47 ∆ mix G / RT = x 1 ln ϕ1 + x 2 ln ϕ 2 + gx 1 ϕ 2

[4.4.12a]

∆ mix G / RTV = ( x 1 / V1 )ln ϕ1 + (x 2 / V2 ) ln ϕ 2 + BRTϕ1 ϕ 2

[4.4.12b]

or

where: xi ϕi g

B V Vi

mole fraction of component i volume fraction of component i integral polymer-solvent interaction function that refers to the interaction of a solvent molecule with a polymer segment, the size of which is defined by the molar volume of the solvent V1 interaction energy-density parameter that does not depend on the definition of a segment but is related to g and the molar volume of a segment Vseg by B = RTg/Vseg molar volume of the mixture, i.e. the binary polymer solution molar volume of component i

The first two terms of Equation [4.4.12] are named combinatorial part of ∆mixG, the third one is then a residual Gibbs free energy of mixing. Applying Equation [4.4.11] to [4.4.12], one obtains:  1 ∆µ 1 / RT = ln(1 − ϕ 2 ) +  1 −  ϕ 2 + χϕ 22  r

[4.4.13a]

  1  χ = ∆µ 1 / RT − ln(1 − ϕ 2 ) −  1 − ϕ 2  / ϕ 22  r  

[4.4.13b]

  1  χ = ln a1 − ln(1 − ϕ 2 ) −  1 −  ϕ 2  / ϕ 22  r  

[4.4.13c]

or

or

where: r

ratio of molar volumes V2/V1, equal to the number of segments if Vseg = V1

4.4 Measurement of solvent activity χ

153

Flory-Huggins interaction function of the solvent

The segment number r is, in general, different from the degree of polymerization or from the number of repeating units of a polymer chain but proportional to it. One should note that Equations [4.4.12 and 4.4.13] can be used on any segmentation basis, i.e., also with r = V2* / V1* on a hard-core volume segmented basis and segment fractions instead of volume fractions, or with r = M2/M1 on the basis of mass fractions. It is very important to keep in mind that the numerical values of the interaction functions g or χ depend on the chosen basis and are different for each different segmentation! From the rules of phenomenological thermodynamics, one obtains the interrelations between both parameters at constant pressure and temperature: χ = g + ϕ1

g=

1 ϕ1

∂g ∂g = g − (1 − ϕ 2 ) ∂ϕ1 ∂ϕ 2

[4.4.14a]

ϕ1

∫ χdϕ

[4.4.14b]

1

0

A recent discussion of the g-function was made by Masegosa et al.50 Unfortunately, gand χ-functions were not always treated in a thermodynamically clear manner in the literature. Sometimes they were considered to be equal, and this is only true in the rare case of composition independence. Sometimes, and this is more dangerous, neglect or misuse of the underlying segmentation basis is formed. Thus, numerical data from literature has to be handled with care (using the correct data from the reviews5-8,11 is therefore recommended). A useful form for their composition dependencies was deduced from lattice theory by Koningsveld and Kleintjens:51 g =α +

β (1 − γϕ 2 )

and

χ =α +

β(1 − γ)

(1 − γϕ 2 )

2

[4.4.15]

where: α β γ

acts as constant within a certain temperature range describes a temperature function like β = β 0 + β1 / T is also a constant within a certain temperature range.

Quite often, simple power series are applied only: i +1 n n  χ   1 − ϕ 2  χ = ∑ χ i ϕ i2 and g = ∑  i    i =0 i = 0  i + 1  1 − ϕ 2 

[4.4.16]

where: χi

empirical fitting parameters to isothermal-isobaric data

Both interaction functions are also functions of temperature and pressure. An empirical form for these dependencies can be formulated according to the rules of phenomenological thermodynamics: g = β 00 + β 01 / T + (β10 + β11 / T ) P or χ = a + b / T + (c + d / T ) P

[4.4.17]

154

Christian Wohlfarth

where: a,b,c,d β 00 , β 01 , β10 , β11 T P

empirical fitting parameters for the χ-function empirical fitting parameters for the g-function absolute temperature pressure

All these fitting parameters may be concentration dependent and may be included into Equations [4.4.15 or 4.4.16]. Details are omitted here. More theoretical approaches will be discussed shortly in Subchapter 4.4.4. The χ-function can be divided into an enthalpic and an entropic parts:  ∂χT   ∂χ  χ = χ H + χ S with χ H = −T    and χ S =   ∂T  P , ϕ  ∂T  p, ϕ

[4.4.18]

where: χH χS

enthalpic part entropic part

An extension of all these equations given above to multi-component mixtures is possible. Reviews of continuous thermodynamics which take into account the polydisperse character of polymers by distribution functions can be found elsewhere.52-54 4.4.3 EXPERIMENTAL METHODS, EQUIPMENT AND DATA REDUCTION 4.4.3.1 Vapor-liquid equilibrium (VLE) measurements Investigations on vapor-liquid equilibrium of polymer solutions are the most important source for obtaining solvent activities in polymer solutions. Therefore, emphasis is laid to the experimental methods which use this equilibrium. These methods are: (i) absolute vapor pressure measurement, (ii) differential vapor pressure measurement, (iii) isopiestic sorption/desorption methods, i.e. gravimetric sorption, piezoelectric sorption, or isothermal distillation, (iv) inverse gas-liquid chromatography (IGC) at infinite dilution, IGC at finite concentrations, and head-space gas-chromatography (HSGC), (v) ebulliometry and (vi) the non-equilibrium steady-state method vapor-pressure osmometry (VPO). The measurement of vapor pressures for polymer solutions is generally more difficult and more time-consuming than that of low-molecular mixtures. The main difficulties can be summarized as follows: Polymer solutions exhibit strong negative deviations from Raoult’s law. These are mainly due to the large entropic contributions caused by the difference between the molar volumes of solvents and polymers, as was explained by the classical Flory-Huggins theory46,47 about 60 years ago, Equation [4.4.12]. However, because of this large difference in molar mass, vapor pressures of dilute solutions do not differ markedly from the vapor pressure of the pure solvent at the same temperature, even at polymer concentrations of 10-20 wt%. This requires special techniques to measure very small differences in solvent activities. Concentrated polymer solutions are characterized by rapidly increasing viscosities with increasing polymer concentration. This leads to a strong increase in time required to obtain real thermodynamic equilibrium caused by a slow solvent diffu-


155

sion effects (in or out of a non-equilibrium-state polymer solution). Furthermore, only the solvent coexists in both phases because polymers do not evaporate. The experimental techniques used for the measurement of vapor pressures of polymer solutions have to take into account all these effects. Vapor pressures of polymer solutions are usually measured in the isothermal mode by static methods. Dynamic methods are seldom applied, see under ebulliometry below. At least, one can consider measurements by VPO to be dynamic methods, where a dynamic (steady-state) balance is obtained. However, limits for the applicable ranges of polymer concentration and polymer molar mass, limits for the solvent vapor pressure and the measuring temperature and some technical restrictions prevent its broader application, see below. Static techniques usually work at constant temperature. The three different methods (i)-(iii) were used to determine most of the vapor pressures of polymer solutions in the literature. All three methods have to solve the problems of establishing real thermodynamic equilibrium between liquid polymer solution and solvent vapor phase, long-time temperature constancy during the experiment, determination of the final polymer concentration and determination of pressure and/or activity. Methods (i) and (ii) were mostly used by early workers. The majority of recent measurements was done with the isopiestic sorption methods. Gas-liquid chromatography as IGC closes the gap at high polymer concentrations where vapor pressures cannot be measured with sufficient accuracy. HSGC can be considered as some combination of absolute vapor pressure measurement with GLC. The following text will now explain some details of experimental equipment and measuring procedures as well as of data reduction procedures to obtain solvent activities. A recent review by Williamson55 provides corresponding information related to low-molecular mixtures. 4.4.3.1.1 Experimental equipment and procedures for VLE-measurements (i) Absolute vapor pressure measurement Absolute vapor pressure measurement may be considered to be the classical technique for our purposes, because one measures directly the vapor pressure above a solution of known polymer concentration. Refs. 56-65 provide a view of the variety of absolute vapor pressure apparatuses developed and used by different authors. The common principle of an absolute vapor pressure apparatus is shown in Figure 4.4.2. Vapor pressure measurement and solution equilibration were made separately: A polymer sample is prepared by weighing, the sample flask is evacuated, degassed solvent is introduced into the flask and the flask is sealed thereafter. All samples are equilibrated at elevated temperature in a thermostat for some weeks (!). The flask with the equilibrated polymer solution is connected to the pressure measuring device (in Figure 4.4.2 a Hg-manometer) at the measuring temperature. The vapor pressure is measured after reaching equilibrium and the final polymer concentration is obtained after correcting for the amount of evaporated solvent. Modern equipment applies electronic pressure sensors and digital techniques to measure the vapor pressure, e.g. Schotsch and Wolf57 or Killmann et al.58 Data processing can be made online using computers. Figure 4.4.3 shows a schematic diagram of the equipment used by Killmann and coworkers.58 A number of problems have to be solved during the experiment. The solution is usually in an amount of some cm3 and may contain about 1 g of polymer or even more. Degassing is absolutely necessary. For example, Killmann et al.58 included special degassing units for each component of the entire equipment. All impurities in the pure solvent have to

156

Figure 4.4.2. Schematic of the common principle of an absolute vapor pressure apparatus: 1 - polymer solution, 2 - connection to the manometer, 3 - Hg-manometer, 4 heating coils. The whole construction is thermostated at the measuring temperature, the connection to the manometer is kept slightly above the measuring temperature to avoid condensation.

Christian Wohlfarth

be eliminated. Equilibration of all prepared solutions is very time-consuming (liquid oligomers need not so much time, of course). Increasing viscosity makes the preparation of concentrated solutions more and more difficult with further increasing amount of polymer. Solutions above 50-60 wt% can hardly be prepared (depending on the solvent/polymer pair under investigation). The determination of the volume of solvent vaporized in the unoccupied space of the apparatus is difficult and can cause serious errors in the determination of the final solvent concentration. To circumvent the vapor phase correction, one can measure the concentration directly by means, for example, of a differential refractometer. The contact of solvent vapor with the Hg-surface in older equipment may cause further errors. Complete thermostating of the whole apparatus is necessary to avoid condensation of solvent vapors at colder spots. Since it is disadvantageous to thermostat Hg-manometers at higher temperatures, null measurement in-

Figure 4.4.3. Schematic diagram of a modern absolute vapor pressure apparatus: T - temperature meter, P - vapor-pressure meter, V - vacuum meter, Z - measuring cell, M - magnetic stirrer, GI and GII - degassing units for the solvent and for the polymer. [Reprinted with permission from Ref. 58, Copyright 1990,Wiley-VCH].


157

struments with pressure compensation were sometimes used, e.g., Baxendale et al.59 Modern electronic pressure sensors can be thermostated within certain temperature ranges. If pressure measurement is made outside the thermostated equilibrium cell, the connecting tubes must be heated slightly above the equilibrium temperature to avoid condensation. The measurement of polymer solutions with lower polymer concentrations requires very precise pressure instruments, because the difference in the pure solvent vapor pressure becomes very small with decreasing amount of polymer. At least, no one can really answer the question if real thermodynamic equilibrium is obtained or only a frozen non-equilibrium state. Non-equilibrium data can be detected from unusual shifts of the χ-function with some experience. Also, some kind of hysteresis in experimental data seems to point to non-equilibrium results. A common consistency test on the basis of the integrated Gibbs-Duhem equation does not work for vapor pressure data of binary polymer solutions because the vapor phase is pure solvent vapor. Thus, absolute vapor pressure measurements need very careful handling, plenty of time, and an experienced experimentator. They are not the method of choice for high-viscous polymer solutions. (ii) Differential vapor pressure measurement The differential method can be compared under some aspects with the absolute method, but there are some advantages. The measuring principle is to obtain the vapor pressure difference between the pure solvent and the polymer solution at the measuring temperature. Figure 4.4.4 explains the basic principle as to how it is used by several authors. References66-75 provide a view of a variety of differential vapor pressure apparatuses developed and used by different authors. The polymer sample is put, after weighing, into the sample flask and the apparatus is evacuated. Degassed solvent is distilled into the measuring burette and from there a desired amount of solvent is distilled into the sample flask. The Hg-manometer is filled from the storage bulb and separates the polymer solution from the burette. Care must be taken to avoid leaving any solvent in the manometer. The apparatus is kept at constant measuring temperature, completely immersed in a thermostat for several days. After reaching equilibrium, the pressure is read from the manometer difference and the concentration is calculated from the calibrated burette meniscus difference corrected by the amount of vaporized solvent in the unoccupied space of the equipment. The pure solvent vapor pressure is usually precisely known from independent experiments. Difference/differential manometers have some advantages from their construction: They are comFigure 4.4.4. Schematic diagram of a differparatively smaller and their resolution is much ential vapor-pressure apparatus: 1 - connection to vacuum pump, 2 - Hg-storage bulb, 3 higher (modern pressure transducers can resolve dif- burette, 4 - Hg-manometer, 5 - polymer so- ferences of 0.1 Pa and less). However, there are the lution. The whole apparatus is kept constant same disadvantages with sample/solution preparaat the measuring temperature within a thertion (solutions of grams of polymer in some cm3 volmostat.

158

Christian Wohlfarth

ume, degassing, viscosity), long-time thermostating of the complete apparatus because of long equilibrium times (increasing with polymer molar mass and concentration/viscosity of the solution), correction of unoccupied vapor space, impurities of the solvent, connection to the Hg-surface in older equipment and there is again the problem of obtaining real thermodynamic equilibrium (or not) as explained above. Modern equipment uses electronic pressure sensors instead of Hg-manometers and digital technique to measure the vapor pressure. Also thermostating is more precise in recent apparatuses. The apparatus developed by Haynes et al.66 is shown in Figure 4.4.5 as example. Problems caused by the determination of the unoccupied vapor space were Figure 4.4.5. Differential vapor-pressure apparatus. 100 avoided by Haynes et al., since they meaml Pyrex flasks connected (a) to a differential pressure transducer (c) with digital readout (d) and (b) to vacuum sure the pressure difference as well as the pump (e) and absolute pressure vacuum thermocouple absolute vapor pressure. Also, the concengauge (f). The constant temperature in the water bath is tration is determined independently by usmaintained by a temperature controller (g). The trans- ing a differential refractometer and a ducer and connecting glassware are housed in an insulated box (i) and kept at constant temperature slightly normalized relation between concentration above the measuring temperature by controller (j). Poly- and refractive index. Degassing of the liqmer solution and pure solvent (here water) are stirred by uids remains a necessity. Time for estabunderwater magnetic stirrers (h). [Reprinted with per- lishing thermodynamic equilibrium could mission from Ref. 66, Copyright 1989, American Chembe somewhat shortened by intensive stirical Society]. ring (slight problems with increasing polymer concentration and solution viscosity were reported). In comparison to absolute vapor-pressure measurements, differential vapor-pressure measurements with a high resolution for the pressure difference can be applied even for dilute polymer solutions where the solvent activity is very near to 1. They need more time than VPO-measurements, however. (iii) Isopiestic sorption/desorption methods Isopiestic measurements allow a direct determination of solvent activity or vapor pressure in polymer solutions by using a reference system (a manometer has not necessarily to be applied). There are two general principles for lowering the solvent activity in the reference system: concentration lowering or temperature lowering. Isopiestic measurements have to obey the condition that no polymer can vaporize (as it might be the case for lower-molecular oligomers at higher temperatures). Concentration lowering under isothermal conditions is the classical isopiestic technique, sometimes also called isothermal distillation. A number of solutions (two as the minimum) are in contact with each other via their common solvent vapor phase and solvent


Figure 4.4.6. Schematic of the experimental arrangement for an isopiestic measurements. [Reprinted with permission from Ref. 77, Copyright 1995, Elsevier Science].

159

evaporates and condenses (this is the isothermal “distillation” process) between them as long as the chemical potential of the solvent is equal in all solutions. At least one solution serves as reference system, i.e., its solvent activity vs. solvent concentration dependence is precisely known. After an exact determination of the solvent concentration in all equilibrated solutions (usually by weighing), the solvent activity in all measured solutions is known from and equal to the activity of the reference solution, Equations [4.4.1 to 4.4.5]. This method is almost exclusively used for aqueous polymer solutions, where salt solutions can be applied as reference systems. It is a standard method for inorganic salt systems. Examples of this technique are given elsewhere.76-80 Figure 4.4.6 provides a scheme of the experimental arrangement for isopiestic measurements as used by Grossmann

et al.77 to illustrate the common principle. The complete apparatus consists of six removable stainless steel cells placed in hexagonal pattern in a copper block. The copper block is mounted in a chamber which is thermostated. Each cell has a volume of about 8 cm3 and is closed by a removable lid. During the experiment, the cells are filled with about 2 cm3 polymer solution (or reference solution) and placed into the copper block. The chamber is sealed , thermostated and evacuated. The lids are then opened and solvent is allowed to equilibrate between the cells as explained above. After equilibration, the cells are closed, removed from the chamber and weighed precisely. Equilibrium requires usually a couple of days up to some weeks. During this time, the temperature of the thermostat does not fluctuate by less than ±0.1 K, which is realized by the copper block that works as a thermal buffer. Temperature lowering at specified isobaric or isochoric conditions is the most often used technique for the determination of solvent vapor pressures or activities in polymer solutions. The majority of all measurements is made using this kind of an isopiestic procedure where the pure solvent is used as reference system. The equilibrium condition of equal chemical potential of the solvent in the polymer solution as well as in the reference system is realized by keeping the pure solvent at a lower temperature (T1) than the measuring temperature (T2) of the solution. At equilibrium, the vapor pressure of the pure solvent at the lower temperature is then equal to the partial pressure of the solvent in the polymer solution, i.e., P1s (T1) = P1(T2). Equilibrium is again established via the common vapor phase for both subsystems. The vapor pressure of the pure solvent is either known from independent data or measured additionally in connection with the apparatus. The composition of the polymer solution can be altered by changing T1 and a wide range of compositions can be studied (between 30-40 and 85-90 wt% polymer, depending on the solvent). Measurements above 85-90 wt% polymer are subject to increasing errors because of surface adsorption effects.

160

Christian Wohlfarth

There is a broad variety of experimental equipment that is based on this procedure (see below). This isopiestic technique is the recommended method for most polymer solutions since it is advantageous in nearly all aspects of measurement: It covers the broadest concentration range. Only very small amounts of polymer are needed (about 30-50 mg with the classical quartz spring balance, about 100 µg with piezoelectric sorption detector or microbalance techniques - see below). It is much more rapid than all other methods explained above, because equilibrium time decreases drastically with such small amounts of polymer and polymer solution (about 12-24 hours for the quartz spring balance, about 3-4 hours for piezoelectric or microbalance techniques). The complete isotherm can be measured using a single loading of the apparatus. Equilibrium is easier to obtain since comparatively small amounts of solvent have to diffuse into the bulk sample solution. Equilibrium can better be tested by measuring sorption and desorption runs which must lead to equal results for thermodynamic absorption equilibrium. Supercritical solvents can be investigated if the piezoelectric detector is used (otherwise buoyancy in dense fluids may cause serious problems). Much broader pressure and temperature ranges can be covered with relatively simple equipment, what may again be limited by the weighing system. Isopiestic sorption measurements can be automated and will allow also kinetic experiments. There are two disadvantages: First, isopiestic sorption measurements below about 30 wt% polymer are subject to increasing error because very small temperature differences (vapor pressure changes) are connected with large changes in concentration. Second, problems may arise with precise thermostating of both the solvent and the solution at different constant temperatures over a longer period of time. Because of their importance, several technical solutions will now be presented in some detail. The classical concept is the sorption method using a quartz spring balance. Refs. 81-90 provide some examples, where the concentration (mass) of the solution is measured by the extension of the quartz spring according to Hook’s law (linear relationship, no hysteresis). It was not originally developed for polymer solutions but for gas-solid adsorption measurements by McBain.91 The principle was introduced into the investigation of polymer solutions by van der Waals and Hermans84 and became popular after the work of Bonner and Prausnitz.85 In this method, a weighed quantity of the (non-volatile) polymer is placed on the pan of the quartz spring balance within a measuring cell. The determination of spring extension vs. mass has to be made in advance as a calibration procedure. Reading of the spring extension is usually made by means of a cathetometer. The cell is sealed, evacuated and thermostated to the measuring temperature (T2), and the solvent is then introduced into the measuring cell as solvent vapor. The solvent vapor is absorbed by the polymer sample to form the polymer solution until thermodynamic equilibrium is reached. The solvent vapor is provided from a reservoir either of pure liquid solvent thermostated at a lower temperature (T1) or of a reference liquid solution of known concentration/solvent partial pressure like in the case of the isothermal distillation procedure as described above. A compact version of such an apparatus was developed by Illig82 and widely used within the author’s own work (see Appendix 4.4A for the corresponding references). Figure 4.4.7a shows the details of the equilibrium cell, which has a vacuum double-walled jacket. The following problems have to be solved during the experiment: The equilibrium cell has to be sealed carefully to avoid any air leakage over the complete duration of the measurements (to measure one isotherm lasts about 14 days). Specially developed thin Teflon sealing rings were preferred to grease. The polymer sample has to withstand the tempera-


161

ture. Changes by thermal ageing during the experiment must be avoided. The temperatures provided by the thermostats must not fluctuate more than ±0.1 K. Condensation of solvent vapor at points that become colder than T2 has to be avoided by slight overheating (this problem may arise at the closing plug and at the connection between the pure solvent reservoir flask and the double-walled jacket). An intelligent improvement of this compact apparatus was made by Sadowski,87 see Figure 4.4.7b, where this vapor-sorption apparatus is combined with technical solutions from ebulliometry (more about Figure 4.4.7a. Isopiestic vapor Figure 4.4.7b. Dynamic isopiestsorption apparatus using a quartz ic vapor-sorption apparatus us- ebulliometers can be found bespring: 1 - connection to the vac- ing a quartz spring (drawing low). uum line, 2 - connection to the provided by G. Sadowski): a) In comparison to the usual thermostating unit which real- evaporator, b) superheater, c) ebulliometric equipment where izes the constant measuring tem- measuring cell, d) condenser, e) perature T2 (the correct value of quartz spring, f) polymer sam- the polymer solution is placed into T2 is obtained by a Pt-100 resis- ple/solution, g) Pt-100 resistance the evaporator, only pure solvent tance thermometer within the thermometer. [Reprinted with is evaporated. The vapor flows cell that is not shown), 3 - closing permission from Ref. 87, Copythrough the cell and is condensed plug, 4 - quartz spring (reading right 1995, Wiley-VCH]. at its head-condenser to flow back of its extension is made by a cathetometer), 5 - sample pan into the reservoir at the bottom. with the polymer solution, 6 - The vapor pressure is kept constant using a manostat and is pure solvent reservoir at tempermeasured additionally outside the apparatus after the conature T1. [Reprinted with permission from Ref. 82, Copyright denser. Equilibrium times decrease somewhat, degassing of 1982, Wiley-VCH]. the solvent is not necessary, air leakage does not play any role. As was stated by different authors, additional measurement of the vapor pressure inside the isopiestic sorption apparatus seems to be necessary if there is some doubt about the real pressure or if no reliable pure solvent vapor pressure data exist for the investigated temperature range. Figure 4.4.8 shows an apparatus used by the author for measurements between room temperature and 70oC and pressures up to 1.5 bar. It combines mercury float valves with Hg-manometers to avoid the use of any grease within the measuring system, a kind of equipment proposed earlier by Ashworth and Everett.88 Up to four quartz springs can be inserted into the equilibrium cell (only one is shown). Reading of the manometer and of the extension of the quartz spring was made using a cathetometer. The direct pressure measurement has the advantage that absolute pressures can be obtained and pressure fluctuations can be observed. More modern equipment applies electronic pressure sensors instead of Hg-manometers to avoid the problems caused by the

162

Christian Wohlfarth

contact of solvent vapor with the mercury surface and to get a better resolution of the measuring pressure. Isopiestic vapor sorption can be made using an electronic microbalance instead of the quartz spring balance. Again, this was not originally developed for polymer solutions but for gas-solid adsorption measurements where this is a widespread application. Electronic microbalances are commercially available from a number of Figure 4.4.8. Isopiestic vapor-sorption apparatus with built-in ma- producers. Their main advantages nometer using a quartz spring: 1 - connection to the vacuum, 2-9 - are their high resolution and their stop corks, 10, 11, 12 - connections to nitrogen, 13 - degassing flask possibility to allow kinetic meafor the pure solvent, 14, 18 - buffers, 15 - cold trap, 16, 19 - Hg-masurements. Additionally, experinometers, 17, 20 - mercury float valves, 21 - pure solvent reservoir at using electronic temperature T1 provided by 22 - thermostat, 23 - temperature con- ments trolled air box, 24 - measuring cell, 25 - quartz spring (four quartz microbalances can easily be autosprings can be inserted into the equilibrium cell, only one is shown), mated and provide computing fa26 - pan with the polymer solution, 27 - closing plug sealed with epcilities. The major disadvantage oxy resin, 28 - heating to avoid solvent condensation. with some of these microbalances is that they cannot be used at high solvent vapor pressures and so are limited to a relatively small concentration range. However, since thin polymer films can be applied, this reduces both the time necessary to attain equilibrium (some hours) and the amount of polymer required and equilibrium solvent absorption can be obtained also at polymer mass fractions approaching 1 (i.e., for small solvent concentrations). Depending on their construction, the balance head is situated inside or outside the measuring apparatus. Problems may arise when it is inside where the solvent vapor may come into contact with some electronic parts. Furthermore, all parts of the balance that are inside the apparatus have to be thermostated to the measuring temperature to enable the correct equilibration of the polymer solution or even slightly above to avoid condensation of solvent vapor in parts of the balance. The allowed temperature range of the balance and its sensitivity to solvent corrosion determine then the accessible measuring range of the complete apparatus. Yoo and coworkers92,93 have recently measured various polymer solutions with such equipment and Figure 4.4.9 shows some details of their apparatus. Two thermostats maintain the pure solvent temperature T1 and the measuring temperature T2 as described above for the spring balance technique, thermostat three protects the essential part of the balance for solvent vapor condensation and damage. A calibrated weight was loaded on the left side of the balance. A granular type of quartz was used as reference weight in order to prevent possible solvent vapor condensation. A dish-type quartz sorption cell was used to load the polymer sample. Platinum wire was used to link both arms to the balance to prevent possible oxidative corrosion of the arm by the solvent. The vapor pressure is measured directly by applying a W-tube Hg-Manometer. The manometer reading was made using a cathetometer.


Figure 4.4.9. Schematic diagram of an isopiestic vapor sorption apparatus using an electronic microbalance: PC - personal computer, MB - microbalance, WB1-3 - water bath thermostats with T3>T2>T1, V1-3 - valves, WM - W-tube mercury manometer, S - polymer sample/solution, SV - solvent reservoir, MS magnetic stirrer, CT - cold trap, VP - vacuum pump. [Reprinted with permission from Ref. 92, Copyright 1998, American Chemical Society].

163

Comparable apparatuses were constructed and used, for example, by Bae et al.94 or by Ashworth,95 Ashworth and Price96 applied a magnetic suspension balance instead of an electronic microbalance. The magnetic suspension technique has the advantage that all sensitive parts of the balance are located outside the measuring cell because the balance and the polymer solution measuring cell are in separate chambers and connected by magnetic coupling only. This allows its application even at very high temperatures of some hundred degrees as well

as pressures up to hundreds of MPa. The most sensitive solvent vapor sorption method is the piezoelectric sorption detector. The amount of solvent vapor absorbed by a polymer is detected by a corresponding change in frequency of a piezoelectric quartz crystal coated with a thin film of the polymer because a frequency change is the response of a mass change at the surface of such a crystal. The frequency of the crystal decreases as mass increases when the crystal is placed in a gas or vapor medium. The frequency decrease is fairly linear. The polymer must be coated onto the crystal from a solution with some care to obtain a fairly uniform film. Measurements can be made at dynamic (vapor flow) or static conditions. With reasonable assumptions for the stability of the crystal’s base frequency and the precision of the frequency counter employed, the piezoelectric method allows the detection of as little as 10 nanograms of solvent using a 10 MHz crystal. This greatly reduces both the time necessary to attain equilibrium (3-4 hours) and the amount of polymer required. Saeki et al.97-99 extensively applied this method to various polymer solutions in a concentration range between 60 and 100 wt% polymer. Recently, Wong et al.100 and Mikkilineni et al.101 presented some new investigations with this method. Figure 4.4.10 shows a schematic diagram of the general equipment. A resolution of nanograms could be realized by Mikkilineni et al.101 Measurements were also made as a function of time to obtain diffusion coefficients. Comparison with gravimetric sorption measurements demonstrated the accuracy of the experiment. Ref.100 presents some details about the electronic circuit, the mounting arrangements for the quartz crystals and the sorption cell. Because very thin films are applied, equilibrium solvent absorption also can be obtained at polymer mass fractions approaching 1 (i.e., for small sol-

164

Christian Wohlfarth

vent concentrations). Sorption- desorption hysteresis has never been observed when using piezoelectric detectors. Bonner and Prausnitz85 reported some hysteresis results when applying their quartz spring sorption balance for polymer concentrations above 85 wt%. This demonstrates the effect of reducing the amount of polymer from about 50 mg for the quartz spring sorption technique by an order of 103 for the piezoelectric detector. However, measurements are limited to solvent concentrations well below the region where solution drops would Figure 4.4.10. Schematic diagram of an isopiestic vapor be formed. On the other hand, measuresorption apparatus using a piezoelectric crystal detector. [Reprinted with permission from Ref. 101, Copyright ments also can be made at higher temperatures and pressures. Limits are set by the 1995, American Chemical Society]. stability of the electrical equipment and the construction of the measuring cell. (iv) Gas-liquid chromatography (GLC) In 1969 Smidsrod and Guillet102 demonstrated that GLC could be used to determine the activity coefficient of a solute in a (molten) polymer at essentially zero solute concentration. This type of activity coefficient is known as an infinite-dilution activity coefficient. Smidsrod and Guillet also introduced the term “inverse” gas-liquid chromatography (IGC) because in IGC the liquid polymer in the stationary phase acts as a solvent for the very small amount of an injected solvent sample like the solute in this case. Methods and results of the application of IGC to polymers and polymer solutions have been reviewed continuously13-25 so that an extensive discussion is not required here. The equipment in principle does not differ very much from that used in analytical GLC. Figure 4.4.11 is a schematic of a simple IGC unit.

Figure 4.4.11. Schematic diagram of an IGC apparatus: 1 - carrier gas, 2 - pressure reducer, 3 - gas cleaning unit (if necessary) , 4+5 - gas-pressure regulation and control unit, 6 - manometer, 7 - column, 8 - thermostat, 9 - mechanical mixer, 10 - inlet syringe, 11 - detector (the gas flows after the detector through a bubble flow meter that is not shown here), 12 - electronics, 13 - recorder.


165

For infinite dilution operation the carrier gas flows directly to the column which is inserted into a thermostated oil bath (to get a more precise temperature control than in a conventional GLC oven). The output of the column is measured with a flame ionization detector or alternately with a thermal conductivity detector. Helium is used today as carrier gas (nitrogen in earlier work). From the difference between the retention time of the injected solvent sample and the retention time of a non-interacting gas (marker gas), the thermodynamic equilibrium behavior can be obtained (equations see below). Most experiments were made up to now with packed columns, but capillary columns were used, too. The experimental conditions must be chosen so that real thermodynamic data can be obtained, i.e., equilibrium bulk absorption conditions. Errors caused by unsuitable gas flow rates, unsuitable polymer loading percentages on the solid support material and support surface effects as well as any interactions between the injected sample and the solid support in packed columns, unsuitable sample size of the injected probes, carrier gas effects, and imprecise knowledge of the real amount of polymer in the column, can be sources of problems, whether data are nominally measured under real thermodynamic equilibrium conditions or not, and have to be eliminated. The sizeable pressure drop through the column must be measured and accounted for. Column preparation is the most difficult task within the IGC-experiment. In the case of packed columns, the preparation technique developed by Munk and coworkers103,104 is preferred, where the solid support is continuously soaked with a predetermined concentration of a polymer solution. In the case of capillary IGC, columns are made by filling a small silica capillary with a predetermined concentration of a degassed polymer solution. The one end is then sealed and vacuum is applied to the other end. As the solvent evaporates, a thin layer of the polymer is laid down on the walls. With carefully prepared capillary surfaces, the right solvent in terms of volatility and wetting characteristics, and an acceptable viscosity in the solution, a very uniform polymer film can be formed, typically 3 to 10 µm thick. Column preparation is the most time-consuming part of an IGC-experiment. In the case of packed columns, two, three or even more columns must be prepared to test the reproducibility of the experimental results and to check any dependence on polymer loading and sometimes to filter out effects caused by the solid support. Next to that, various tests regarding solvent sample size and carrier gas flow rate have to be done to find out correct experimental conditions. There is an additional condition for obtaining real thermodynamic equilibrium data that is caused by the nature of the polymer sample. Synthetic polymers are usually amorphous or semi-crystalline products. VLE-based solvent activity coefficients require the polymer to be in a molten state, however. This means that IGC-measurements have to be performed for our purpose well above the glass transition temperature of the amorphous polymer or even above the melting temperature of the crystalline parts of a polymer sample. On the other hand IGC can be applied to determine these temperatures. The glass transition of a polymer does not take place at a fixed temperature but within a certain temperature range depending on the probing technique applied because it is a non-equilibrium effect. Figure 4.4.12 demonstrates the appearance of the glass transition region in an IGC-experiment. The S-shaped part of the curves in Figure 4.4.12 is the glass transition region. Its minimum describes the glass transition temperature as obtained by IGC. Only data from the straight line on the left side at temperatures well above the glass transition temperature lead

166

Christian Wohlfarth

to real thermodynamic vapor-liquid equilibrium data. As a rule of thumb, the experimental temperature must exceed the glass transition temperature by about 50K. Form and width of the S-shaped region depend somewhat on the solvent used and, as can be seen from the picture, there is a certain dependence on molar mass of the polymer. Data on the right side at temperatures below the glass transition describe mainly surface adsorption effects. Brockmeier et al.106,107 showed that GLC can also be used to determine the partial pressure of a solute in a polymer solution at concentrations as great as 50 wt% solute. In this case of finite concentration Figure 4.4.12. Temperature dependence of the specific retention volume Vg of p-xylene in polystyrenes of vary- IGC, a uniform background concentration ing molar masses, experimental data were measured by of the solute is established in the carrier gas. Glindemann.105 The carrier gas is diverted to a saturator through a metering valve. In the saturator it passes through a diffuser in a well-stirred, temperature-controlled liquid bath. It leaves the separator with the solute equilibrium vapor pressure in the carrier gas. The solute concentration is varied by changing the saturator temperature. Precise control of the temperature bath is needed in order to obtain a constant plateau concentration. Upon leaving the saturator the gas flows to the injector block and then to the column. As in the infinite dilute case a small pulse of the solvent is then injected. This technique is known as elution on a plateau, Conder and Purnell.108,109 Because finite concentration IGC is technically more complicated, only few workers have applied it. Price and Guillet110 demonstrated that results for solvent activity, activity coefficient or χ-function are in good agreement with those obtained by traditional isopiestic vapor sorption methods. Whereas the vapor sorption results are more accurate at higher concentrations, the reverse is true for finite concentration IGC since larger injection volumes have to be used, which strains the theory on which the calculations are based. Also, at large vapor concentrations the chromatographic peaks become more spread out, making the measurement of retention times less precise. Additionally, the concentration range is limited by the requirement that the saturator temperature must be below that of the column. Clearly, at higher measuring temperatures, higher solvent concentrations may be used. Finite concentration IGC can be extended to multi-component systems. Especially ternary polymer solutions were investigated to some extend with this technique, e.g., Bonner and coworkers111,112 or Glover and coworkers.113-115 Data reduction is somewhat complicated, however. Danner et al.116 tested the frontal analysis by characteristic point (FACP) technique to measure thermodynamic data for polymer-solvent systems at finite concentrations. In the FACP technique, a complete isotherm can be derived from the shape of one breakthrough profile. A point on an isotherm is obtained by measuring the retention volume of the characteristic point at the corresponding concentration. The methods to determine thermodynamic data by FACP technique were discussed in detail by Conder and Young.117


167

Wolf and coworkers118,119 applied another GLC technique to VLE-measurements for polymer-solvent systems, the so-called head-space gas-chromatography (HSGC). This is practically a combination of static vapor pressure measurement with gas-chromatographic detection. HSGC experiments were carried out with an apparatus consisting of a head-space-sampler and Figure 4.4.13. Schematic of applying head-space a normal gas chromatograph. Figure 4.4.13 gas-chromatography (HSGC) to VLE-measurements in shows a schematic diagram of the equippolymer solutions (drawing provided by B. A. Wolf, ment. Univ. Mainz, Germany). The pneumatically driven thermostated headspace-sampler samples a constant amount of gas phase (that must be in equilibrium with the liquid polymer solution, of course) and injects this mixture into the gas chromatograph. Helium is used as carrier gas. After separation of the components of the gaseous mixture in a capillary column they are detected individually by a thermal conductivity detector. The signals are sent to an integrator which calculates the peak areas, which are proportional to the amount of gas in the sample volume and consequently to the vapor pressure. Calibration can be made by measuring the pure solvent in dependence on temperature and to compare the data with the corresponding vapor pressure vs. temperature data. Measurements can be done between about 25 and 85 wt% polymer in the solution (again depending on temperature, solvent and polymer investigated). In order to guarantee thermodynamic equilibrium, solutions have to be conditioned for at least 24 h at constant temperature in the head-space-sampler before measurement. Degassing is not necessary and solvents have to be purified only to the extent necessary to prevent unfavorable interactions in the solution. The experimental error in the vapor pressures is typically of the order of 1-3%. Details about theory and practice of HSGC were discussed by Kolb and Ettre.120 One great advantage of HSGC is its capability to measure VLE-data, not only for binary polymer solutions but also for polymer solutions in mixed solvents, since it provides a complete analysis of the vapor phase in equilibrium. This is usually not the case with the classical isopiestic sorption balances where PVT-data and a material-balance calculation must be included into the data reduction to calculate vapor phase concentrations, e.g., Refs.121-123 (v) Ebulliometry (boiling point elevation of the solvent) As pointed out above, dynamic vapor-liquid equilibrium measurement methods are not very suitable for concentrated polymer solutions, especially due to their heavy foaming behavior. For dilute polymer solutions, however, there is continuing application of ebulliometry as an absolute method for the direct determination of the number-average molecular mass Mn. Dedicated differential ebulliometers allow the determination of values up to an order of 100,000 g/mol. Ebulliometry as a method for molar mass determination was recently reviewed by Cooper,33 Glover,34 and Mays and Hadjichristidis.40 The major requirements for a successful ebulliometry experiment are thermal stability, equilibration of both concentration and temperature, temperature measurement and control and pressure measurement and control. It is an advantage of ebulliometry to know very exactly the constant pressure applied since pressure constancy is a prerequisite of any

168

Christian Wohlfarth

successful experiment. Commercially sold ebulliometers have seldom been used for polymer solutions. For application to polymer solutions, the operating systems have been individually constructed. The above-mentioned reviews explain some of these in detail which will not be repeated here as ebulliometry is not really a practiced method to obtain solvent activities and thermodynamic data in polymer solutions. However, ebulliometry is a basic method for the investigation of vapor-liquid equilibrium data of common binary liquid mixtures, and we again point to the review by Williamson,55 where an additional number of equilibrium stills is shown. Ebulliometers have traditionally been classified as either simple, in which only a single temperature is measured, or differential, in which the boiling temperatures of the pure solvent and of the solution were measured simultaneously. In differential ebulliometers, two independent temperature sensors can work, or a single differential temperature measurement is done. Essentially, all ebulliometers for polymer solutions are of the differential type. The manner in which the reference boiling temperature of the pure solvent is provided differs, however. Establishment and maintenance of both temperature and concentration equilibrium are accomplished in a variety of ways. The common method is the use of a vapor lift pump (a Cottrell pump) where the boiling liquid is raised to a position from which it can flow in a thin film until superheat is dissipated and its true boiling temperature can be measured. This technique has one disadvantage: the pumping rate depends on the heat input. This is of particular importance with polymer solutions in which problems due to foaming occur. To overcome this problem mechanical pumps were sometimes applied. Other ebulliometer types have been reported that use the methods of surface volatilization, spray cooling, two-stage heating, or rotating ebulliometer; for more details please see Refs.33,34,40 Methods of temperature measurement within ebulliometer experiments will not be discussed here, as they change rapidly with continuing progress of electronics and computerization. Pressure control is important for single temperature ebulliometers, as the boiling temperature depends on pressure. It is not so important in differential type ebulliometers, owing to the simultaneous and compensating change in reference temperatures. Therefore, direct changes in boiling temperatures present no serious problem if sufficient time is allowed for calibration. It is usually recommended that the ebulliometer be thoroughly cleaned and dried between experiments. Small amounts of polymer adsorbed on the surface must be avoided. (vi) Vapor-pressure osmometry (VPO) Vapor-pressure osmometry is, from its name, compared with membrane osmometry by considering the vapor phase to act like the semipermeable membrane, however, from its principles it is based on vapor pressure lowering or boiling temperature elevation. Since the direct measure of vapor pressure lowering of dilute polymer solutions is impractical because of the extreme sensitivity that is required, VPO is in widespread use for oligomer solutions (Mn less than 20,000 g/mol) by employing the thermoelectric method as developed by Hill in 1930.124 In the thermoelectric method, two matched temperature-sensitive thermistors are placed in a chamber that is thermostated to the measuring temperature and where the atmosphere is saturated with solvent vapor. If drops of pure solvent are placed on both thermistors, the thermistors will be at the same temperature (zero point calibration). If a solution drop is placed on one thermistor, a temperature difference ∆T occurs which is caused by condensation of solvent vapor onto the solution drop. From equilibrium thermodynamics follows that this temperature increase has its theoretical limit when the vapor pressure of


169

the solution is equal to that of the pure solvent, i.e., at infinite dilution. The obtained temperature difference is very small, about 10-5K. Because solvent transfer effects are measured, VPO is a dynamic method. This leads to a time-dependent measurement of ∆T. The principle scheme of a VPO apparatus is given in Figure 4.4.14. Today, vapor-pressure osmometers are Figure 4.4.14. Principle scheme of a vapor-pressure osmometer of the hanging commercially available drop type. T - measuring temperature, ∆T - obtained temperature difference (time dependent), measurements are made at atmospheric pressure where T de- from a number of protermines the partial vapor pressure of the solvent P1 in air. ducers. They can be divided into two basic types: those that employ conventional hanging drop thermistors as in Figure 4.4.14 and those that use vertical thermistors. The vertical thermistors automatically control drop size to ensure more reproducible response. The hanging drop design requires the operator to manually monitor and control drop size. Furthermore, commercial instruments have been developed which utilize vertical thermistors having cups or pieces of platinum gauze to control drop size in a highly reproducible manner. More details about instrumentation and techniques can be found in the reviews given by Glover,34 Mays and Hadjichristidis.40 A very recent presentation can be found in a new book edited by Pethrick and Dawkin.26 Depending on technical details of the equipment, on the sensitivity of the temperature detector, on measuring temperature, solvent vapor pressure and polymer concentration in the solution drop, a steady state for ∆T can be obtained after some minutes. The value of ∆Tst is the basis for thermodynamic data reduction (see below). If measuring conditions do not allow a steady state, an extrapolation method to ∆T at zero measuring time can be employed for data reduction. Sometimes a value is used that is obtained after a predetermined time; however, this may lead to some problems with knowing the exact polymer concentration in the solution. The extrapolation method is somewhat more complicated and needs experience of the experimentator but gives an exact value of polymer concentration. Both methods are used within solvent activity measurements when polymer concentrations are higher and condensation is faster than in common polymer characterization experiments. A way to avoid these problems is discussed below. Experience has shown that careful selection of solvent and temperature is critical to the success of the VPO experiment. Nearly all common solvents, including water (usually, there are different thermistor sensors for organic solvents and for water), can be used with VPO. The measuring temperature should be chosen so that the vapor pressure of the solvent will be greater than 6,000 Pa, but not so high as to lead to problems with evaporation from the chamber. Solvent purity is critical, especially volatile impurities, and water must be

170

Christian Wohlfarth

avoided. Greater sensitivity can be achieved by using solvents with low enthalpies of vaporization. This means, for our task, that not all desirable polymer-solvent pairs and not all temperature (pressure) ranges can be investigated by VPO. Additionally, VPO has some inherent sources of error. These belong to the possible existence of surface films, to differences in diffusion coefficients in solutions, to appreciably different solution concentrations, to differences in heat conductivity, to problems with drop size and shape, to the occurrence of reactions in the solution, and to the presence of volatile solutes. Of course, most of them can be avoided by laboratory practice and/or technical improvements, but it must be taken into account when measuring solvent activities. Regener and Wohlfarth125 developed a way to enlarge the applicability range of VPO to polymer concentrations ≤40wt% for the purpose of measuring solvent activities. An increase of polymer concentration over the linear steady state working range of VPO causes some problems. First, no thermodynamically defined ∆T can be obtained and, second, the calibration constant may become dependent on concentration. Thus, the only way to achieve higher concentrations is to find methods to minimize the increasing chemical potential difference of the solvent between the two drops. This can be achieved by using a reference solution of known solvent activity instead of the pure solvent. The instrument is then used as a zero-point detector comparing the solvent activity of the reference solution with solvent activity of the polymer solution. The reference concentration has to be varied until ∆T = 0 is found. The only assumption involved in this method is equal solvent condensation and diffusion. The extrapolation method to ∆T at zero measuring time can be used to minimize these influences. It is not really necessary to find the reference solution at exactly ∆T=0, but it is sufficient to measure a small ∆T < 0 and small ∆T > 0 and to interpolate between both known solvent activities. An example is shown in Figure 4.4.15, where benzene was used as solute for the reference solutions. Since the polymer solution remains quasi unchanged in concentration, this modified VPO-method is faster than isopiestic isothermal distillation experiments with organic solvents and polymer solutions. Difficulties with the increasing viscosity of concentrated polymer solutions set limits to its applicability, because solutions should flow easily to form drops. Recently, Gaube et al.126,127 or Eliassi 128 et al. measured water activities in aqueous solutions of poly(ethylene glycol) and showed that the conventional VPO method also can be used for higher polymer concentrations with good success. 4.4.3.1.2 Primary data reduction Equation [4.4.7] is the starting relation for data from VLE-measurements. Two relations are necessary to obtain the solvent acFigure 4.4.15. Experimental data of the system toluene + tivity a1: one for the fugacity coefficient of polystyrene, Mn = 1380 g/mol, at 323.15K, isopiestic vapor pressure/sorption measurement (full circles), VPO at the solvent vapor and one for the standard higher concentrations (gray circles), data from authors state fugacity of the liquid solvent. In prinown work. ciple, every kind of equation of state can be


171

applied to calculate the solvent vapor fugacity coefficient. This is done if dedicated equations of state are applied for further modeling. However, in most cases it is common practice to use the virial equation of state for the purpose of reducing primary VLE-data of polymer solutions. This procedure is sufficient for vapor pressures in the low or medium pressure region where most of the VLE-measurements are performed. The virial equation is truncated usually after the second virial coefficient, and one obtains from Equation [4.4.6]:  P   RT 

m m  m ln φi =  2∑ y i B ij − ∑ ∑ y i y j B ij i =1 j =1  j =1

[4.4.19]

where: Bii Bjj Bij

second virial coefficient of pure component i at temperature T second virial coefficient of pure component j at temperature T second virial coefficient corresponding to i-j interactions at temperature T.

In the case of a strictly binary polymer solutions Equation [4.4.19] reduces simply to: ln φ1 =

B11P RT

[4.4.20]

To calculate the standard state fugacity, we consider the pure solvent at temperature T and saturation vapor pressure Ps for being the standard conditions. The standard state fugacity is then calculated as:

(

)

V1L P − P1s + B11P1s f = P exp  RT  s 1

0 1

  

[4.4.21]

where: P1s V1L

saturation vapor pressure of the pure liquid solvent 1 at temperature T molar volume of the pure liquid solvent 1 at temperature T

The so-called Poynting correction takes into account the difference between the chemical potentials of the pure liquid solvent at pressure P and at saturation pressure P1s assuming that the liquid molar volume does not vary with pressure. Combining Equations [4.4.7, 4.4.20 and 4.4.21] one obtains the following relations:

(

a1 = φ y 1P / f = P1 / P V 1

0 1

s 1

)

(

)(

 B11 − V1L P − P1s exp  RT 

(

)(

 B11 − V1L P − P1s γ 1 = a1 / x 1L = P1 / x 1L P1s exp  RT 

(

)

)  

)  

[4.4.22a]

[4.4.22b]

172

Christian Wohlfarth

(

)(

 B11 − V1L P − P1s Ω 1 = a1 / w1L = P1 / w1L P1s exp  RT 

(

)

)  

[4.4.22c]

These relations can be applied to VLE-data from all experimental methods. The data reduction for infinite dilution IGC starts with the usually obtained terms of retention volume or net retention volume. Vnet = Vr − Vdead

[4.4.23]

where: Vnet Vr Vdead

net retention volume retention volume retention volume of the (inert) marker gas, dead retention, gas holdup

These retention volumes are reduced to specific ones by division of Equation [4.4.23] with the mass of the liquid (here the liquid, molten polymer), corrected for the pressure difference between column inlet and outlet pressure and reduced to T0 = 273.15K. V T Vg0 =  net   0  m 2  T

 3(Pin / Pout ) − 1  3  2(Pin / Pout ) − 1 2

[4.4.24]

where: Vg0 m2 Pin Pout

specific retention volume corrected to 0oC mass of the polymer in the liquid phase within the column column inlet pressure column outlet pressure

Theory of GLC provides the relation between Vg0 and thermodynamic data for the low-molecular component (solvent) 1 at infinite dilution:  P1  xL  1

∞

 TR  = 0  Vg0 M 2 

or

 P1   wL  1

∞

 TR  = 0  Vg0 M1 

[4.4.25]

where: M2 M1

molar mass of the liquid (molten) polymer molar mass of the low-molecular component (solvent).

The activity coefficients at infinite dilution follow immediately from Equation [4.4.22] by introducing the above result, if we neglect interactions to and between carrier gas molecules (which is normally helium):  TR γ 1∞ =  0 0 s V M P  g 2 1

(

)(

 B − V1L P − P1s   exp  11  RT  

)  

[4.4.26a]


 TR Ω 1∞ =  0 0 s V M P  g 2 1

173

(

P s V L − B11   exp  1 1  RT  

)  

[4.4.26b]

The standard state pressure P has to be specified. It is common practice by many authors to define here zero pressure as standard pressure since pressures are usually very low during GLC-measurements. Then, Equations [4.4.26a and b] change to:  TR γ 1∞ =  0 0 s V M P  g 2 1

P s V L − B11   exp  1 1  RT  

) 

[4.4.27a]

 TR Ω = 0 0 s V M P  g 1 1

P s V L − B11   exp  1 1  RT  

(

) 

[4.4.27b]

∞ 1

(





One should keep in mind that mole fraction-based activity coefficients become very small values for common polymer solutions and reach the value of 0 for M2→ ∞, which means a limited applicability to at least oligomer solutions. Therefore, the common literature provides only mass fraction-based activity coefficients for (high-molecular) polymer/(low-molecular) solvent pairs. Furthermore, the molar mass M2 of the polymeric liquid is an average value according to the usual molar-mass distribution of polymers. Additionally, it is a second average if mixed stationary liquid phases are applied. Furthermore, thermodynamic VLE-data from GLC-measurements are provided in the literature as values for (P1/w1)∞, see Equation [4.4.25], i.e., classical mass fraction based Henry’s constants (if assuming ideal gas phase behavior): H1, 2

P =  1L  w1

∞

 TR  = 0  Vg0 M1 

[4.4.28]

Thus, Equation (4.4.27b) reduces to Ω 1∞ =

(

P1s V1L − B11  exp RT P1s 

H1, 2

)  

[4.4.29]

The data reduction for finite concentration IGC by elution on a plateau is more complicated than for infinite dilution IGC via Equations [4.4.24 to 26] and will not be explained here. A detailed analysis of the elution on plateau mode was made by Conder and Purnell.108,109 For the determination of thermodynamic properties of polymer solutions by finite-concentration IGC the reader is referred to the paper by Price and Guillet110 who provide a comprehensive derivation of all necessary equations. The data reduction of ebulliometric measurements can be made either by using Equations [4.4.22] or by applying the relation for the boiling point elevation of a binary mixture:

174

Christian Wohlfarth

∆T ebull = −

RT 2 ln a1 ∆vap H10

[4.4.30]

where: T ∆Tebull ∆ vap H10

measuring temperature (= boiling point temperature of the pure solvent) temperature difference of boiling point elevation molar enthalpy of vaporization of the pure solvent 1 at temperature T.

The ratio M1RT2/∆ vap H10 is called the ebulliometric constant. For the determination of solvent activities from ebulliometric data, tabulated ebulliometric constants should not be used, however. On the other side, it is sometimes recommended to use reference solutes to establish an experimental relationship for the equipment in use, i.e., unprecise data for the enthalpy of vaporization or perhaps some non-equilibrium effects cancel out of the calculation. Enthalpies of vaporization are provided by several data collections, e.g., by Majer and Svoboda,129 or through the DIPPR database.130 The data reduction of vapor-pressure osmometry (VPO) follows to some extent the same relations as outlined above. However, from its basic principles, it is not an equilibrium method, since one measures the (very) small difference between the boiling point temperatures of the pure solvent drop and the polymer solution drop in a dynamic regime. This temperature difference is the starting point for determining solvent activities. There is an analogy to the boiling point elevation in thermodynamic equilibrium. Therefore, in the steady state period of the experiment, the following relation can be applied if one assumes that the steady state is sufficiently near the vapor-liquid equilibrium and linear non-equilibrium thermodynamics is valid: ∆T st = −kVPO

RT 2 ln a1 ∆vap H10

[4.4.31]

where: T ∆Tst kVPO ∆ vap H10

measuring temperature (= temperature of the pure solvent drop) temperature difference between solution and solvent drops in the steady state VPO-specific constant molar enthalpy of vaporization of the pure solvent 1 at temperature T .

Recent examples of solvent activity measurements by VPO in aqueous solutions of poly(ethylene glycol) by Eliassi et al.128 and of poly(ethylene glycol) or dextran by Gaube et al.126,127 demonstrate the obtainable high quality if precise experiments were made. The so-called VPO-specific constant contains all deviations from equilibrium state and it is to be determined experimentally. It depends on certain technical details from the equipment used and also on the temperature and solvent applied. It is assumed not to depend on the special solute under investigation and can therefore be obtained by calibration. Equation [4.4.31] can also be used if not the steady state, but the temperature difference extrapolated to a measuring time of zero is determined by the experimentator. However, the values of kVPO are different for both methods. A more detailed discussion about calibration problems can be found in the papers of Bersted,131,132 or Figini.133-135 Usually, VPO-data are reduced to virial coefficients and not to solvent activities. Power series expansion of Equation [4.4.31] leads to the following relations:


175

 1  + Γ2 c 2 + Γ3 c 22 +K  M  2 

[4.4.32a]

 RT 2V1  1 ∆T st = kVPO + Γ2′ c ′2 + Γ3′ c ′22 +K 0  c ′2 ∆ LV H1 M 2 

[4.4.32b]

RT 2V1 ∆T st = kVPO c2 ∆ LV H10 or

where: c2 c′2 ν mi V1 M1 M2 Γ 2,Γ 3, ... Γ2′ , Γ3′, ..

mass by volume concentration c2 = m2/ν mass by mass concentration c2′ = m2/m1 volume of the polymer solution mass of component i molar volume of the solvent molar mass of the solvent molar mass of the polymer second, third, ... VPO-virial coefficients based on g/cm3 concentrations second, third, ... VPO-virial coefficients based on g/g concentrations

In the dilute concentration region, these virial equations are usually truncated after the second virial coefficient which leads to a linear relationship. These truncated forms of Equation [4.4.32] are the basis for applying VPO to polymer characterization, which will not be discussed here - please see Refs.26,34,40 Solvent activities can be estimated from second virial coefficients with some care regarding the necessary accuracy of all numerical values included. The molar mass of the polymer, M2, is the number-average, Mn, if polydisperse samples are investigated. Corresponding averages of the virial coefficients can be introduced, too. The estimation of higher virial coefficients than the second one is difficult and hardly leads to satisfying results, because measurements at high polymer concentrations cause a lot of problems. In some cases, however, as in the above-mentioned paper by Gaube et al.,126,127 precise measurements were done for polymer concentrations up to 30-40 wt% and second and third virial coefficients were obtained in good quality. As pointed out above, there is another way VPO can be applied to measure activity differences between two polymer solution drops that differ slightly in concentration (in the same solvent, of course). In this case, VPO is quasi an isopiestic experiment and the unknown activity can be determined by using reference solutions with known solvent activity values:125 a1(T, wpolymer) = a1(T, wreference)

[4.4.33]

Reference solutions can be made with the same organic solutes that are used for calibration. In the case of water, NaCl or KCl solutions may be applied as it is done for many isopiestic (isothermal distillation) measurements with aqueous solutions. 4.4.3.1.3 Comparison of experimental VLE-methods The general aim of all experiments is to measure solvent activities in polymer solutions over the complete concentration range and for all desired temperatures (and pressures). Additionally, the dependence on molar mass of the polymer has to be taken into account. As is clear from all explanations above, there is no really universal method to fulfill all purposes.

176

Christian Wohlfarth

Vapor pressure/vapor sorption measurements cover nearly the complete concentration range if different apparatuses are used. Measurements can be made with good accuracy. Principal limits with respect to temperature or pressure do not exist, but most apparatuses in the literature are constructed only for temperatures between 20 and 100oC, sometimes up to 150oC, and pressures between 1 and 100 - 200 kPa. Vapor pressure/vapor sorption measurements are very time-consuming experiments. To obtain a complete isotherm one needs usually about a month with conventional techniques or, at least, some days with microbalances or piezoelectric sensors. This demands long-time stability for thermostating and precise temperature control. Furthermore, the equilibrium cell has to be sealed in such a way that air leakage is avoided for the complete duration of the measurement. Experimentators need quite a lot of experience until they observe really good data. Experiments can only partially be automated depending on the method and equipment applied. The accuracy of the final data depends on the method applied, the temperature or pressure investigated, and also the given concentration. Measurements above about 85 wt% polymer showed sometimes sorption-desorption hysteresis. Solvent degassing is absolutely necessary with the exception of the apparatus proposed by Sadowski where degassing takes place automatically during the experiment (see above). The solvent must be purified from all other impurities. This is true of course also for the polymer investigated. According to their capabilities, different apparatuses should be used: differential pressure measurements for 5-30 wt% polymer in the solution, isopiestic sorption techniques for 30-85 wt% polymer, piezoelectric or microbalance detection for 60-99 wt% polymer. These limits can change somewhat with molar mass of the polymer. Oligomer solutions are easier to handle and can be measured even with conventional VLE-technique as developed for low-molecular liquid mixtures. There may be limits in temperature and pressure that depend on the nature of the solvent/polymer pair. Usually, the solutions investigated should not show liquid-liquid demixing and solutions should not become solid. Thermodynamic equilibrium data can only be obtained if the polymer is investigated well above its glass transition temperature. There is a depression of the glass transition temperature with increasing solvent concentration, but there are polymers that can be investigated only at temperatures above 100oC and more, even in concentrated solutions. VPO is more limited with respect to the measurement of solvent activities. It is designed only for dilute polymer solutions (in the maximum up to 40 wt% polymer), optimum temperature and pressure (well below normal pressure) ranges and molar masses up to about 20,000 g/mol for the polymer. Not all solvents can be applied. On the other hand, VPO is a well-established method, commercially available, possessing a high resolution for very small differences of solvent activities with respect to the pure solvent and does not need much time. Steady-state conditions are obtained within minutes and quite a lot of measurements can be made during a working day. There are no problems with external thermostating or long-time stability. Experimental results from VPO are in good agreement with measurements from scattering techniques. VPO measurements close the gap between 0 and 30 wt% polymer in the solution with respect to conventional vapor pressure/vapor sorption measurements (of course, only within its limits explained). Experimentators easily acquire the necessary experience with the measuring equipment. The piezoelectric sorption technique is a method that is especially suitable for the low solvent concentration range. It is the most sensitive solvent vapor sorption method. A resolution of nanograms can be realized. Measurements can also be made as a function of time


177

to obtain diffusion coefficients. Comparison with gravimetric sorption measurements demonstrated the accuracy of the experiment. Because very thin films are applied, equilibrium solvent absorption also can be obtained at polymer mass fractions approaching 1, as with the IGC experiment. Comparison to IGC-data gives good agreement. Sorption-desorption hysteresis has never been observed when using piezoelectric detectors. Measurements are limited to a concentration range where the swollen polymer film is still stable at the crystal surface. Equilibrium is rather quickly established, usually after 3-4 hours, i.e., an isotherm can be measured within some days. With the corresponding equipment, high pressures and high temperatures can be applied, too. IGC is the most rapid method and it is the recommended technique for the infinite dilution range of the solvent in the (liquid, molten) polymer. Measurements can also be made to obtain diffusion coefficients. Column preparation and finding optimum experimental conditions are the most time-consuming tasks. These tasks require quite a lot of experience. The final measurements can be automated and provide quick, reliable and reproducible results. Temperature and solvent dependencies can easily be investigated. The common accuracy is 1-3% with respect to data of the χ-function or Henry’s constant. There is no need to degas the solvents or to purify them except from impurities which may react with the polymer. Limits are mainly given by the glass transition temperature of the polymer as explained above. Due to this problem, most IGC measurements are made at temperatures well above 100oC. On the other hand, temperatures well above 100oC can cause the problem of thermal ageing and degradation of the polymer sample if temperatures are too high. In comparison to IGC, vapor pressure measurements were made in most cases below 100oC. There were some special investigations in earlier literature to compare IGC-data at infinite dilution with those from vapor pressure measurements at concentrated solutions, e.g., Refs.110,136-138 Differences between IGC-data and vapor pressure measurements reported in older papers are mainly caused by errors with the IGC technique. Temperatures were used too near or even within the glass transition region, unsuitable polymer loading was applied, non-equilibrium conditions were used. But, there are also errors from/within vapor pressure data, mainly sorption/desorption hysteresis at too high polymer concentrations because of non-equilibrium conditions. Today it is accepted that there are no differences between IGC-data and vapor pressure measurements if all thermodynamic equilibrium conditions are carefully obeyed. In contrast to vapor pressure measurements, IGC can also be applied with thermodynamically bad solvents. It is the only method to obtain limiting activity coefficients for strong non-solvents. Even mass fraction based activity coefficients above 25 or χ-values of 2 or more can be measured. Finite concentration IGC provides the possibility to connect advantages from IGC and vapor pressure measurements because it can be applied between 50 and 100 wt% polymer. However, the experimental technique is more sophisticated, data reduction is more complicated, and only few workers have applied it. On the other hand, much experimental time can be saved since finite concentration IGC is a rapid method. One isotherm can be observed within one day (or two). Price and Guillet110 or Danner et al.116 demonstrated that results for solvent activity coefficients and χ-functions or sorption isotherms are in good agreement with those obtained by traditional isopiestic vapor sorption methods. The concentration range of finite concentration IGC is limited by the requirement that the saturator temperature must be below that of the column. Clearly, at higher measuring temperatures, higher

178

Christian Wohlfarth

solvent concentrations may be used. Finite concentration IGC can be extended to multi-component systems. Head-space gas chromatography is a modern tool for the measurement of vapor pressures in polymer solutions that is highly automated. Solutions need time to equilibrate, as is the case for all vapor pressure measurements. After equilibration of the solutions, quite a lot of data can be measured continuously with reliable precision. Solvent degassing is not necessary. Measurements require some experience with the equipment to obtain really thermodynamic equilibrium data. Calibration of the equipment with pure solvent vapor pressures may be necessary. HSGC can easily be extended to multi-component mixtures because it determines all components in the vapor phase separately. In summary, the decision for a special equipment depends to some extend on concentration, temperature and pressure ranges one is interested in. From the experience of the author, the combination of isopiestic vapor pressure/vapor sorption measurements for the determination of solvent activities with infinite dilution IGC for the determination of Henry’s constants provides good experimental data and covers a temperature range that is broad enough to have a sufficient data basis for thermodynamic modeling. If one is interested in both solvent solubility and diffusion data, finite concentration IGC or piezoelectric sorption techniques should be applied. 4.4.3.2 Other measurement methods This subchapter summarizes all other experimental methods mentioned in subchapter 4.4.1 in order of their special importance and use regarding the determination of solvent activities in polymer solutions. 4.4.3.2.1 Membrane osmometry Apart from VLE-measurements, membrane osmometry is the next important method that has been used for measuring solvent activities in polymer solutions. This follows from the tables in Refs.1,2,5,8 according to its occurrence in comparison to the other methods. Most of these measurements were made in the dilute solution regime; only a small number of papers dealt with high-pressure osmometry where one also can measure solvent activities for concentrated solutions with polymer concentrations up to about 50 wt%, e.g. Refs.139-145 Laboratory designed instruments were developed in the 40’s and 50’s, e.g. by Zimm145 or by Flory.144 Later on, high speed membrane osmometers are commercially available, e.g., from Knauer, HewlettPackard or Wescan Instruments. External pressures may be applied to balance the osFigure 4.4.16. Principle scheme of a membrane motic pressure if necessary, e.g., Vink.140 osmometer: 1 - solvent, 2 - polymer, π - osmotic pressure, The principle scheme of a membrane ∆h - hydrostatic height difference, P0 - ordinary pressure or measuring pressure, V - partial molar volume of the osmometer together with the corresponding 1

solvent in the polymer solution.


179

thermodynamic situation is illustrated in Figure 4.4.16. Technical details of the different apparatuses will not be presented here, however, the principle construction of the measuring cell and the heating thermostat of the Knauer membrane osmometer A 300 is shown in Figure 4.4.17 for illustration and as example. As a general feature of most osmometers, the membrane is clamped into a stainless steel thermostated chamber (the measuring cell and the pressure measuring system of modern osmometers are built into a high-grade electronically stabilized thermostat) and serves as barrier between the pure solvent and the polymer solution sides of the chamber. The solvent side (bottom) is in juxtaposition with a pressure sensor, e.g. the diaFigure 4.4.17. Description of the principal construction of a Knauer membrane osmometer A 300: 1 - head thermostat, 2 - channel for sy- phragm of a capacitance strain ringe, 3 - calibration device with suction tube, 4 - calibration glass, 5 gauge or a piezo-chip. The solvent - capillary position MEASUREMENT, 6 - capillary position transport is measured across the CALIBRATION, 7 - tension screws, 8 - cell retaining disc, 9 - upper bottom side of the membrane in half of measuring cell, 10 - sample introduction system, 11 semipermeable membrane, 12 - lower half of measuring cell, 13 - the direction of the solution which pressure measuring system, 14 - cell thermostat, 15 - suction of cali- is topside the membrane. The bration bottle. [Reprinted from the operating manual with permis- amount of flowing solvent is in the sion from Dr. H. Knauer GmbH (Germany)]. range of 10-6 ml and equilibrium is established after some minutes when hydrostatic pressure prevents further solvent flow. This is indicated by the electronics of the equipment as well as any changes in equilibrium such as thermal drift of solute diffusion through the membrane. Other osmometers apply compensation methods where the increase of the hydrostatic height of the solution side is automatically compensated by changing the filling height. Due to this procedure, only very small amounts of solvent have to permeate through the membrane and equilibrium is reached within 10-20 minutes. The classical procedure was only used in older laboratory designed instruments where one started at zero and measured the hydrostatic height difference as a function of time until equilibrium is reached. More details about instrumentation and techniques can be found in the reviews by Adams,38 Tombs and Peacock,39 Mays and Hadjichristidis.40 A very recent presentation can be found in a new book edited by Pethrick and Dawkins.26

180

Christian Wohlfarth

Some efforts are necessary to keep the osmometer under appropriate working conditions. This relates mainly to the proper preconditioning and installation of the membrane, the attainment of thermal equilibrium, the calibration of the electronic output, the adjustment of solvent zero, and to choosing the desired sensitivity. For aqueous solutions, cellulose acetate membranes are usually employed, but any dialysis or ultrafiltration or reverse osmosis membrane can be used, too. The membranes should be conditioned in solvent or buffer and degassed before use while still in the solvent. For organic solvents, gel cellulose or cellophane membranes are preferred. They must be conditioned to a new solvent by gradual changes of the corresponding solvent mixture. Details are usually given by the supplier. Membranes in various pore sizes are recommended for solutes of low molar mass. Aging and deswelling of membranes lead to decreasing permeability and increasing measuring times. Adsorption of polymer molecules at the membrane surface, “ballooning” of the membrane due to unfavorable pressure effects, membrane asymmetry and action of surface active substances on the membrane must be avoided. The relation between osmotic pressure and solvent activity is to be found from the chemical potential equilibrium condition, taking into account the pressure dependence of µ1. From the rules of phenomenological thermodynamics, one obtains:  ∂µ 10   ∂P 

  = V1  T

and

V − V1  ∂ ln a1    = 1 RT  ∂P T

[4.4.34]

where: V1 V1

partial molar volume of the solvent in the polymer solution at temperature T molar volume of the pure solvent at temperature T

Integration is performed between P0, i.e., the ordinary pressure or measuring pressure, and π, the osmotic pressure, and results in Equation [4.4.35]. ∆µ 1 = RT ln a1 = −V1 π

[4.4.35]

Usually, the experimental data are reduced to virial coefficients and not to solvent activities. Series expansion of Equation [4.4.35] leads to the following relation:  1  π = RT  + A2 c 2 + A3 c 22 +K c2 M 2 

[4.4-36]

where: c2 mass by volume concentration c2 = m2/ν A2,A3, ... second, third, ... osmotic virial coefficients

The molar mass of the polymer, M2, is the number-average, Mn, if polydisperse samples are investigated. Corresponding averages of the virial coefficients can be introduced, too. In the dilute concentration region, the virial equation is usually truncated after the second virial coefficient which leads to a linear relationship. A linearized relation over a wider concentration range can be constructed, Equation [4.4.38], if the Stockmayer-Casassa relation,146 Equation [4.4.37], between A2 and A3 is applied:


A3 M n = (A2 M n / 2)  π  c2

  

0. 5

 RT   =   Mn 

0. 5

2

 A2 M n  1 + 2 c 2   

181

[4.4.37]

[4.4.38]

Examples for experimentally determined virial coefficients can be found in the above mentioned papers139-145 and in the tables prepared by Lechner et al.9 Solvent activities can be calculated via Equations [4.4.35 to 38] from osmotic second virial coefficients with some care regarding the necessary accuracy of all numerical values included. The partial molar volume of the solvent can be approximated in most cases by the molar volume of the pure solvent. Noda et al.139 published a combined investigation of the thermodynamic behavior of poly(α-methylstyrene)s having sharp molar mass distributions and covering a wide range of molar masses in toluene. They applied osmotic pressure, light scattering and vapor pressure measurements and demonstrated the capabilities of these methods in comprehensive and detailed form. Gaube et al.126,127 could show that in the case of aqueous dextran solutions, water activity data and virial coefficients measured by VPO and by membrane osmometry are in good agreement. 4.4.3.2.2 Light scattering Light scattering is one of the most widespread characterization techniques for polymers. Therefore, technical and methodical details will not be explained here - please see Refs.26-32 for such information. The general set-up of a scattering principle is illustrated by Figure 4.4.18. The scattering vector q is the difference between the wave vectors ki and ks of the incident and the scattered plane waves, the scattering angle θ is the angle between both vectors. Both are related by |q| = (4 π / λ 0 )sin(θ / 2), where λ 0 is the wavelength of light in vacuum. Laser light is used today for the light source. Light scattering in homogeneous fluFigure 4.4.18. General set-up of a scattering experiment: ids is caused by fluctuations in the dielectric ki, ks - wave vectors of the incident and the scattered constant. In pure liquids these are due to plane waves, q - scattering (or wave) vector, D- detector, density fluctuations, in homogeneous soluS - sample, θ - scattering angle from the transmitted beam, I0 - incident intensity of unpolarized light, r - the tions mainly to concentration fluctuations distance between sample and detector. which generally lead to much larger fluctuations in dielectric constant than density variations. The difference between solution and pure solvent is called excess scattering. This excess scattering is of interest here, since it is related to the second derivative of Gibbs free energy of mixing with respect to concentration (and via this way to solvent activities): I excess ∝

TP(θ) ∂ 2 ∆ mix G ∂ϕ i ∂ϕ j

[4.4.39]

182

Christian Wohlfarth

where: Iexcess T ∆ mixG ϕi P(θ) θ

excess scattering intensity absolute temperature Gibbs free energy of mixing. volume fractions of component i (= ith polymer species in the molecular distribution) properly averaged particle scattering factor scattering angle from the transmitted beam

The determinant in the denominator is to be calculated at constant temperature and pressure. It reduces to the single second derivative ( ∂ 2 ∆ mix G / ∂ 2 ϕ 2 )P,T = ( ∂µ 1 / ∂ϕ 2 )P,T for the case of a strictly binary monodisperse polymer solution. The average particle scattering factor is of primary importance in studies of the size and shape of the macromolecules, but it is merely a constant for thermodynamic considerations. Conventionally, the so-called Rayleigh factor (or ratio) is applied: R(θ) ≡

I excess r 2

(

I 0V0 1 + cos 2 θ

[4.4.40]

)

where: R(θ) I0 r2 V0

Rayleigh factor incident intensity of unpolarized light square of the distance between sample and detector detected scattering volume

and, neglecting P(θ), Equations [4.4.39 and 4.4.40] can be transformed to: R(θ) ≡

RTKϕ 2V1 (∂µ 1 / ∂ϕ 2 ) P ,T

[4.4.41]

where: V1 ϕ2 K

partial molar volume of the solvent in the polymer solution at temperature T volume fraction of the monodisperse polymer optical constant

The optical constant for unpolarized light summarizes the optical parameters of the experiment:

(

K ≡ 2π 2 n o2 (dn / dc 2 ) P ,T / NAv λ40 2

)

[4.4.42]

where: no n c2 NAv λ0

refractive index of the pure solvent refractive index of the solution mass by volume concentration c2 = m2/ν Avogadro’s number wavelength of light in vacuum

For dilute polymer solutions, the partial derivative in Equation [4.4.41] is a weak function of composition and the scattering intensity increases roughly proportional to the volume fraction of the polymer. While Equation [4.4.41] permits any light scattering data to be


183

interpreted as partial derivatives of solvent chemical potential or activity, dilute solution measurements are conventionally again presented in terms of the osmotic virial expansion:

(Kc 2 ) / R (θ = 0) = (1 / M 2 ) + 2A2 c 2

+ 3A3 c 22 +K

[4.4.43]

where: A2,A3, ... second, third, ... osmotic virial coefficients M2 molar mass of the polymer

According to the scattering theory of polydisperse polymers (please see, for example, the book written by Strobl),147 the molar mass of the polymer, M2, is equal to the mass average, Mw, of polydisperse polymers. The exact application of Equation [4.4.43] is at the scattering angle θ = 0. Interpretation of scattering data at larger angles has to take into account the interparticle interference and the angular variation of the excess scattering intensity. Usually, such data have been analyzed with a Zimm plot,148 where (Kc2)/R(θ) is graphed as a function of sin2(θ/2) + const.*c2 and measurements are extrapolated at constant angle to zero concentration values and at constant concentration to zero angle values. Connecting each of these two sets of points gives the curve specified by Equation [4.4.41]. The intercept gives Mw, the slope of the zero angle data yields the second virial coefficient. In many cases, non-linear Zimm plots were observed.31 An illustrative example for the variation of the second virial coefficient with molar mass and temperature from endothermic to exothermic conditions is given in the paper by Wolf and Adam,149 or in the paper by Lechner and Schulz150 for the variation of the second virial coefficient on pressure, both obtained by light scattering. A recent example for a light scattering investigation on the molar mass dependence of A2 and A3 was published by Nakamura et al.151 The virial expansion is inappropriate as the polymer concentration increases. In these cases, scattering data can be analyzed in terms of a thermodynamic ansatz for the Gibbs free energy of mixing. For example, Scholte152 analyzed light scattering data of concentrated polystyrene solutions by means of the Flory-Huggins approach and determined χ-data. But recently, Hasse, et al.153 combined laser-light scattering with isopiestic measurements and obtained second and third virial coefficients of aqueous poly(ethylene glycol) solutions with high accuracy. Both virial coefficients could be correlated over a temperature range between 278 and 313K, including a derived theta-temperature of 375.7K in good agreement with results from liquid-liquid equilibrium. Additional measurements using membrane osmometry agreed well with the results of the simultaneous correlation of light scattering and isopiestic data. Corresponding measurements of aqueous dextran solutions by Kany et al.154 showed again the resources inherent in such a combination of different methods. Light scattering provides another interesting tool to determine thermodynamic data of polymer solutions. Starting from Equation [4.4.39], Scholte155,156 developed the idea of measuring spinodal curves, i.e., the border between metastable and unstable liquid-liquid demixing behavior of polymer solutions. At this spinodal curve, the determinant in Equation [4.4.39] vanishes, i.e., it becomes equal to zero. At small enough scattering angles (near 30o) and in a temperature range of about 0.03 < ∆T < 5K around the critical temperature, a proportionality of Iexcess ∝1/∆T can be obtained that leads to a simple linear behavior of 1/I30 against T for various concentrations around the critical demixing concentration. The extrapolated curves of 1/I30 vs. T to 1/I30 = 0 lead to spinodal temperatures as function of the corre-

184

Christian Wohlfarth

sponding polymer concentrations. If, for example, one combines ∆mixG-ansatz by Koningsveld and Kleintjens,51 Equation [4.4.15], with the spinodal condition: ∂ 2 ∆ mix G =0 ∂ϕ i ∂ϕ j

[4.4.44]

one obtains for a binary polymer solution: α+

β(1 − γ)

(1 − γϕ 2 )

3

−

1 1 − =0 ϕ 2 rw 1 − ϕ 2

[4.4.45]

where: α β γ rw ϕ2

T

spinodal

acts as constant within a certain temperature range describes a temperature function like β = β0 + β1/T is also a constant within a certain temperature range. mass average segment number, compare r in Equation (4.4.13) total volume fraction of the polymer

The adjustable parameters α, β, γ have to be fitted to spinodal data ϕ spinodal vs. 2 and solvent activities can be calculated from the following relation:

β(1 − γ)  1 ln a1 = ln(1 − ϕ 2 ) +  1 −  ϕ 2 + αϕ 22 + ϕ 22 2 r − 1 γϕ  n  ( 2)

[4.4.46]

where: rn

number average segment number, compare r in Equation [4.4.13]

Gordon and coworkers157-159 improved this method and developed the so-called PICS (pulse-induced critical scattering) apparatus - details and history were summarized by Galina et al.160 PICS enables not only investigations within the metastable range, i.e., nearer to the spinodal, but also of high-viscous solutions and polymer blends for determining spinodal and binodal (cloud-point) curves. How to obtain solvent activities from demixing equilibrium is explained in the text below. 4.4.3.2.3 X-ray scattering X-ray scattering can be measured by the classical Kratky camera or more modern synchrotron techniques. Technical details can be found in a number of books, e.g., those by Guinier and Fouret,161 Chen and Yip,162 or Glatter and Kratky.163 Small angle X-ray scattering (SAXS) can be used in analogy to light scattering to measure second virial coefficients of binary polymer solutions. Zimm-diagrams can be constructed following the same ways as in light scattering. This was demonstrated, for example, in papers by Kirste and coworkers.164-166 In analogy to Equation [4.4.43], one can derive

(Kc 2 ) / I (θ = 0) = (1 / M 2 ) + 2A2 c 2

+ 3A3 c 22 +K

where: c2

mass by volume concentration c2 = m2/ν

[4.4.47]


185

A2,A3, ... second, third, ... osmotic virial coefficients M2 molar mass of the polymer

According to the scattering theory of polydisperse polymers (please see, for example, the book written by Strobl),147 the molar mass of the polymer, M2, is equal to the mass average, Mw, of polydisperse polymers. The exact application of Equation [4.4.47] is to be made again at the scattering angle θ = 0. The constant K is now given by:

(

K ≡ e 4 ∆z 2 / m 2 NAv c 4

)

[4.4.48]

where: e m ∆z c

electron charge electron mass excess number of electrons speed of light in vacuum

In principle, there is agreement between values of second virial coefficients from light scattering or X-ray scattering. Okano et al.167,168 applied SAXS to semidilute solutions of polystyrene in cyclohexane in the poor solvent regime and obtained virial coefficients in good agreement with liquid-liquid data from a coexistence curve. Takada et al.169 provided a more recent example for poly(vinyl methyl ether) in cyclohexane, Horkay et al.170 for poly(vinyl acetate) in toluene and poly(dimethyl siloxane) in octane. In comparison to data from osmotic pressure and neutron scattering, they observed good agreement. 4.4.3.2.4 Neutron scattering Neutron scattering is an important method for investigating conformation and dynamics of polymer molecules, Higgins,171 or polymer mixtures, Hammouda.172 A recent presentation of various techniques can be found in a new book edited by Pethrick and Dawkins.26 Thermodynamics of polymer solutions is not the first task in neutron scattering experiments. The general set-up of the neutron scattering experiment is equivalent to the one used for light scattering, but applying a neutron source, and elastic neutron scattering at small angles (SANS) can be applied like light scattering or X-ray scattering to obtain second virial coefficients in dilute solutions. Similarly to the scattering of photons, it is the difference in scattering power between solvent molecules and polymer segments which determines the absolute scattering intensity. Formally, the virial equation has the same form as Equations [4.4.43 and 47], again neglecting P(θ):

(Kc 2 ) / Σ (θ) = (1 / M 2 ) + 2A2 c 2

+ 3A3 c 22 +K

[4.4.49]

and

(

)

K = b 2 − b 1ρ 1ν 2, spez / NAv where: c2 A2,A3, ... M2 Σ(θ) K b1,b2 ρ1

mass by volume concentration c2 = m2/ν second, third, ... osmotic virial coefficients molar mass of the polymer differential scattering cross section per volume unit contrast factor for neutron scattering densities of solvent and polymer scattering length density of the solvent

[4.4.50]

186

Christian Wohlfarth ν2,spez

specific volume of the polymer (more exact, ν2∞,spez , the partial specific volume at infinite dilution)

According to the scattering theory of polydisperse polymers, please see, for example, in the book written by Strobl,147 the molar mass of the polymer, M2, is equal to the mass average, Mw, of polydisperse polymers. The contrast factor for neutron scattering takes into account for the difference in scattering power of solvent molecules and polymer segments. Again, Zimm plots can be constructed, as was explained above, for light scattering measurements to take into account for angular and concentration dependence - for a demonstration see Vennemann et al.173 The transformation of the obtained second virial coefficients into solvent activities is as explained above, Equations [4.4.34 and 4.4.35]. A recent example for the determination of second virial coefficients from SANS is the investigation of aggregation phenomenon in associating polymer solutions by Pedley et al.,174 where sodium sulfonated polystyrene ionomers in deuterated xylene were considered. Enthalpic and entropic contributions to A2 were calculated (as in the paper by Wolf and Adams149 for A2 from light scattering) and an enthalpy of aggregation was estimated from these data. A high-pressure investigation on aqueous poly(ethylene oxide) solutions was made by Vennemann et al.173 who measured second virial coefficients by a SANS experiment for pressures up to 200 MPa and combined these data with PVT-measurements to obtain also excess and partial excess volumes and gained information about the pressure dependence of the chemical potential. 4.4.3.2.5 Ultracentrifuge The analytical ultracentrifuge is a powerful tool for polymer characterization. Technical details of ultracentrifugation will not be considered here - please see Refs.26,35-37 for more information. In a typical ultracentrifuge experiment, the polymer solution is put in a sample tube and rotated at high speed. Thermodynamic data can be obtained either from the sedimentation velocity (sedimentation coefficient) or from the sedimentation-diffusion equilibrium since the centrifugal forces are balanced by the activity gradient. The concentration gradient is conventionally measured via the refractive index gradient along the axis of the tube using Schlieren photography or various optics. The sedimentation coefficient is defined as the sedimentation velocity in a unit force field: s=

dh / dt ω2 h

[4.4.51]

where: s h t ω

sedimentation coefficient distance from the center of rotation time angular velocity

For a given polymer-solvent system, the sedimentation coefficient is dependent on temperature, pressure and polymer concentration. For obtaining thermodynamic data from sedimentation coefficients, one additionally has to measure the diffusion coefficient. This can be made with an ultracentrifuge in special diffusion cells35 or with dynamic light scattering32 based on the theory of Pecora.175 Nearly all diffusion coefficients have been measured by this method since it became available in 1970. The determination of sedimen-


187

tation and diffusion coefficient yields virial coefficients of a polymer solution. The so-called Svedberg equation reads:  1  D + 2A2 c 2 + 3A3 c 22 +K   1 − ν2, spez ρ 1 = RT  s  M 2 

(

)

[4.4.52]

where: D ν2,spez ρ1 c2 A2,A3, ... M2

diffusion coefficient partial specific volume of the polymer density of the solvent mass by volume concentration c2 = m2/ν second, third, ... osmotic virial coefficients molar mass of the polymer

Equation [4.4.52] is strictly valid for monodisperse polymers, i.e., one single component 2. For polydisperse polymers, different averages were obtained for the sedimentation and the diffusion coefficient, which depends on the applied measuring mode and the subsequent calculations. The averages of M2 correspond with averages of D and s and are mixed ones that have to be transformed into the desired common averages - for details please see Refs.35-37 Sedimentation-diffusion equilibrium in an ultracentrifuge gives also a virial series:35   1  ∂ ln c 2  ω2 h 1 − ν2, spez ρ 1  + 2A2 c 2 + 3A3 c 22 +K  = RT   ∂h   M 2

(

)

[4.4.53]

where: h ω

distance from the center of rotation angular velocity

Equation [4.4.53] is again valid for monodisperse polymers only. Polydisperse polymers lead to apparent molar mass averages and to averages of the virial coefficients which have to be transformed into the desired common averages by appropriate calculation methods.35-37 A somewhat different way of avoiding the virial expansion in Equation [4.4.53] was developed by Scholte.176,177 Without going into details, his final relation was:  M w   ∂∆µ 1 M M   dn   dn + RT  1 − 1    /  ω2 h 1 − ν2, spez ρ 1  1 2  =   M n Mw   dh   dw 2  w 2 − 1  ∂w 2

(

)

  [4.4.54] 

where: w2 M1 Mn Mw n

mass fraction of the polymer molar mass of the solvent number average molar mass of the polymer mass average molar mass of the polymer refractive index of the solution

Some assumptions were made for the derivation of Equation [4.4.54], especially the partial specific volume, the refractive index, and the derivative dn/dw2 must not depend on the molar mass distribution of the polymer. If one further assumes that the Flory-Huggins

188

Christian Wohlfarth

χ-function depends only on temperature and concentration, but not on molar mass, the partial derivative of the chemical potential can be calculated by Equation [4.4.13a] to obtain values of the χ-function. Scholte carried out experiments for solutions of polystyrene in cyclohexane or toluene at different temperatures and in a concentration range of 0-80 wt%. Thus, the sedimentation method is able to cover nearly the total concentration range of a polymer solution; however, values obtained by this method were slightly higher than values determined by other methods. Since the measurement of thermodynamic data by sedimentation equilibrium is not very frequent in the literature this is certainly not a final statement. A combined determination of second osmotic virial coefficients of poly(ethylene glycol)s in methanol, water and N,N-dimethylformamide by Elias and Lys178 using light scattering, VPO and sedimentation equilibrium showed good agreement between all methods. This was also confirmed in a recent investigation on poly(1-phenyl-1-propene) in toluene by Hirao et al.,179 where second virial coefficients were determined by light scattering and by sedimentation equilibrium over a wide range of molar mass. Some further A2 data from sedimentation measurements can be found in the tables by Lechner et al.9 The transformation of the obtained second virial coefficients into solvent activities is as explained above. 4.4.3.2.6 Cryoscopy (freezing point depression of the solvent) In the cryoscopic method, the freezing temperature of a solution is compared with that of the pure solvent. The polymer must be solvable in the solvent at the freezing temperature and must not react with the solvent either chemically or physically. Difficulties may arise from limited solubility and from the formation of solid solutions on freezing. Application of cryoscopy to polymer solutions is not widespread in literature despite the simplicity of the required equipment. Cryoscopy was reviewed by Glover,34 who also discussed technical details and problems in concern with application to polymer solutions. A detailed review on cryometers and cryoscopic measurements for low-molar mass systems was recently made by Doucet.180 Cryometers are sold commercially, e.g., Knauer. Measurements of thermodynamic data are infrequent. Applications usually determine molar masses. Accurate data require precise temperature measurement and control as well as caution with the initiation of the crystallization process and the subsequent establishment of equilibrium (or steady state) conditions. High purity is required for the solvent and also for the solute. Data reduction of cryoscopic measurements is made by applying the relation for the freezing point depression of a binary mixture to obtain solvent activities: 1 1 R − SL = ln a1 0 T1 T1 ∆SL H10

SL

[4.4.55]

where: SL

T10 T1 ∆ SLH10 SL

solid-liquid equilibrium melting temperature of the pure solvent solid-liquid equilibrium melting temperature of the solvent in the polymer solution molar enthalpy of fusion of the pure solvent.

Kawai181 determined some values of the χ-function for benzene solutions of polystyrene or poly(vinyl acetate) and aqueous solutions of poly(vinyl alcohol). In comparison with various data from the tables given by Orwoll,8 larger deviations with respect to other methods have to be stated. Just recently, Hoei et al.182 made a more sophisticated analysis of


189

solid-liquid coexistence data of benzene in natural rubber and found good agreement to other data. Equation [4.4.55] could in principle also be used for the determination of thermodynamic data from the melting point depression of (semi)crystalline polymers when the subscripts changed from 1 to 2. This enables a second approach to data for the infinite dilution range of the solvent in the polymer. Such investigations have been made in the literature. However, these data are regarded as being less reliable by a number of reasons and no further discussion will be made here. 4.4.3.2.7 Liquid-liquid equilibrium (LLE) There are two different situations for the liquid-liquid equilibrium in polymer-solvent systems: (i) the equilibrium between a dilute polymer solution (sol) and a polymer-rich solution (gel) and (ii) the equilibrium between the pure solvent and a swollen polymer network (gel). Case (i) is considered now, case (ii) is specially considered below as swelling equilibrium. LLE-measurements do not provide a direct result with respect to solvent activities. Equation (4.4.8) says that solvent activities at given temperature and pressure must be equal in both coexisting phases. Since the solvent activity of such a coexisting phase is a priori not known, one has to apply thermodynamic models to fit LLE-data as functions of temperature and concentration. Solvent activities can be obtained from the model in a subsequent step only.

Figure 4.4.19. Principles of liquid-liquid demixing in polymer solutions, a) - strictly binary polymer solution of a monodisperse polymer, b) - quasi-binary polymer solution of a polydisperse polymer which is characterized by a distribution function: C - critical point, dashed lines - tie lines, T(1) - temperature/concentration in the homogeneous region, T(2) - temperature/concentrations of the cloud point (phase´) and the corresponding shadow point (phase´´), T(3) - temperature in the heterogeneous LLE region, coexistence concentrations of phase´ and phase´´ at T(3) are related to the starting concentration = cloud point concentration of (2).

190

Christian Wohlfarth

Furthermore, there is another effect which causes serious problems with LLE-data of polymer solutions. This is the strong influence of distribution functions on LLE, because fractionation occurs during demixing - see, for example, Koningsveld.44,183 Figure 4.4.19 illustrates the differences between the LLE-behavior of a strictly binary polymer solution of a monodisperse polymer and a quasi-binary polymer solution of a polydisperse polymer which is characterized by a distribution function. One can see the very complicated behavior of quasi-binary solutions where the phase boundary is given by a cloud-point curve and where an infinite number of coexistence curves exists (one pair for each starting concentration, i.e., each cloud-point). The cloud-point is a point in the T-w2- or the P-w2-diagram where a homogeneous solution of concentration w02 begins to demix (where the “first” droplet of the second phase occurs, T(2) in Figure 4.4.19). If w02 is smaller than the critical concentration, the cloud-point belongs to the sol-phase, otherwise to the gel-phase. As this subchapter is devoted to solvent activities, only the monodisperse case will be taken into account here. However, the user has to be aware of the fact that most LLE-data were measured with polydisperse polymers. How to handle LLE-results of polydisperse polymers is the task of continuous thermodynamics, Refs.52-54 Nevertheless, also solutions of monodisperse polymers or copolymers show a strong dependence of LLE on molar mass of the polymer,184 or on chemical composition of a copolymer.185 The strong dependence on molar mass can be explained in principle within the simple Flory-Huggins χ-function approach, please see Equation [4.4.61]. Experimental methods can be divided into measurements of cloud-point curves, of real coexistence data, of critical points and of spinodal curves: Due to distinct changes in a number of physical properties at the phase transition border, quite a lot of methods can be used to determine cloud-points. In many cases, the refractive index change is determined because refractive indices depend on concentration (with the seldom exception of isorefractive phases) and the sample becomes cloudy when the highly dispersed droplets of the second phase appear at the beginning of phase separation. Simple experiments observe cloud-points visually. More sophisticated equipment applies laser techniques, e.g., Kuwahara,186 and light scattering, e.g., Koningsveld and Staverman.187 The principle scheme of such a scattering experiment is the same as explained with Figure 4.4.18. Changes in scattering pattern or intensity were recorded as a function of decreasing/increasing temperature or pressure. The point, where first deviations from a basic line are detected, is the cloud-point. Since demixing or phase homogenization need some time (especially for highly viscous solutions), special care is to be applied for good data. Around the critical point large fluctuations occur (critical opalescence) and scattering data have to be measured at 90o scattering angle. The determination of the critical point is to be made by independent methods (see below). Various other physical properties have been applied for detecting liquid-liquid phase separation: viscosity, e.g., Wolf and Sezen,188 ultrasonic absorption, e.g., Alfrey and Schneider,189 thermal expansion, e.g., Greer and Jacobs,190 dielectric constant, e.g., Jacobs and Greer,191 or differential thermal analysis DTA, e.g., Muessig and Wochnowski.192 There are only a small number of investigations where real coexistence data were measured. This is mainly due to very long equilibrium times (usually weeks) which are necessary for obtaining thermodynamically correct data. A common method is to cool homogeneous solutions in ampullae very slowly to the desired temperature in the LLE-region and


191

equilibrium is reached after both phases are sharply separated and clear, Rehage et al.193 After separating both phases, concentrations and distribution functions were measured. Acceptable results can be obtained for low polymer concentrations (up to 20 wt%). Scholte and Koningsveld194 developed a method for highly viscous polymer solutions at higher concentrations by constructing a modified ultracentrifuge where the equilibrium is quickly established during cooling by action of gravitational forces. After some hours, concentrations, phase volume ratios and concentration differences can be determined. Rietfeld with his low-speed centrifuge195 and Gordon with a centrifugal homogenizer196 improved this technique and expanded its applicability up to polymer melts, e.g., Koningsveld et al.197 The methods for obtaining spinodal data have already been discussed above with the light scattering technique, please see Subchapter 4.4.3.2.2. Special methods are necessary to measure the critical point. For solutions of monodisperse polymers, it is the maximum of the binodal. Binodals of polymer solutions can be rather broad and flat. The exact position of the critical point can be obtained by the method of the rectilinear diameter. Due to universality of critical behavior, a relation like Equation [4.4.56] is valid, deGennes:198

(ϕ

I 2

(

)

− ϕ II2 / 2 − ϕ crit ∝ 1 − T / T crit 2

)

1− α

[4.4.56]

where: ϕ I2 ϕ II2 ϕ crit 2 Tcrit α

volume fraction of the polymer in coexisting phase I volume fraction of the polymer in coexisting phase II volume fraction of the polymer at the critical point critical temperature critical exponent

and critical points can be obtained by using regression methods to fit LLE-data to Equation [4.4.56]. For solutions of polydisperse polymers, such a procedure cannot be used because the critical concentration must be known in advance to measure its corresponding coexistence curve. Additionally, the critical point is not the maximum in this case but a point at the right-hand side shoulder of the cloud-point curve. Two different methods were developed to solve this problem, the phase-volume-ratio method, e.g., Koningsveld,199 where one uses the fact that this ratio is exactly equal to one only at the critical point, and the coexistence concentration plot, e.g. Wolf,200 where an isoplethal diagram of values of ϕI2 and ϕII2 vs. ϕ 02 gives the critical point as the intersection point of cloud-point and shadow curves. Since LLE-measurements do not provide a direct result with respect to solvent activities, Equation [4.4.8] and the stability conditions are the starting points of data reduction. As pointed out above, the following explanations are reduced to the strictly binary solution of a monodisperse polymer. The thermodynamic stability condition with respect to demixing is given for this case by (see Prausnitz et al. 49):

(∂

2

∆ mix G / ∂ 2 ϕ 2

)

P ,T

>0

[4.4.57]

If this condition is not fulfilled between some concentrations ϕI2 and ϕII2 , demixing is obtained and the minimum of the Gibbs free energy of mixing between both concentrations is given by the double tangent at the corresponding curve of ∆mixG vs. ϕ 2 , Equation [4.4.8].

192

Christian Wohlfarth

The resulting curve in the T vs. ϕ 2 diagram, see Figure 4.4.19a, is the binodal curve. Applying Equations [4.4.3 to 4.4.5 and 4.4.13a], one gets two relations which have to be solved simultaneously to fit an empirical χ(T)-function to experimental binodal (coexistence) data. For the most simple case of χ being only a function of T (or P) and not of ϕ 2 (the so-called one-parameter approach) these relations read: ln

(1− ϕ ) +  1− 1 ϕ ( (1− ϕ )  r  I 2

II 2

I 2

)

(

)

− ϕ II2 + χ ϕ 2I 2 − ϕ 2II 2 = 0

[4.4.58a]

and I 1 ϕ 2  1 I ln II −  1 −  ϕ 2 − ϕ II2 + χ ϕ1I 2 − ϕ1II 2 = 0 r ϕ2  r 

(

(

)

)

[4.4.58b]

where: ϕ I2 ϕ II2 r χ

volume fraction of the polymer in coexisting phase I volume fraction of the polymer in coexisting phase II ratio of molar volumes V2/V1 what is the number of segments with Vseg = V1 Flory-Huggins interaction function of the solvent

and solvent activities result from Equation [4.4.13a]. However, this simple approach is of limited quality. More sophisticated models have to be applied to improve calculation results. A special curve is obtained with the border line to the instability region, i.e., the spinodal curve, for which the second derivative in Equation [4.4.57] is equal to zero. If one applies again the one-parameter approach with an empirical χ(T)-function, the following simple result can be derived:

(

2χ T spinodal

) − rϕ

1

spinodal 2

−

1 spinodal 2

1− ϕ

=0

[4.4.59]

where: Tspinodal ϕ spinodal 2

spinodal temperature volume fraction of the polymer at the spinodal curve

which has to fit an empirical χ(T)-function. An example for the spinodal relation of a polydisperse polymer was given above by Equations [4.4.44 and 4.4.45]. The common point of spinodal and binodal curve is the critical point. The critical point conditions are: (∂ 2 ∆ mix G / ∂ 2 ϕ 2 ) P ,T = 0 ( ∂ 3 ∆ mix G / ∂ 3 ϕ 2 ) P ,T = 0 and

[4.4.60] ( ∂ 4 ∆ mix G / ∂ 4 ϕ 2 ) P ,T > 0

If one applies again the one-parameter approach with an empirical χ(T)-function, two simple results can be derived:


(

ϕ crit = 1 / 1 + r 0. 5 2

)

and

193

(

)

(

χ crit T crit = 0.5 1 + 1 / r 0. 5

)

2

[4.4.61]

where: ϕ crit 2 Tcrit

volume fraction of the polymer at the critical point critical temperature

This means that the critical concentration is shifted to lower values with increasing segment number (molar mass) of the polymer and becomes zero for infinite molar mass. Equation [4.4.61] explains also why the χ(T)-function becomes 0.5 for infinite molar mass. The critical temperature of these conditions is then called theta-temperature. Solvent activities can be calculated from critical χ(T)-function data via Equation [4.4.13]. However, results are in most cases of approximate quality only. 4.4.3.2.8 Swelling equilibrium Polymers, crosslinked sufficiently to form a three-dimensional network, can absorb solvent up to a large extent. The maximum possible solvent concentration, the degree of swelling, are a function of solvent activity. If solvent is present in excess, this swelling equilibrium is reached when the chemical potential of the pure solvent outside the network is equal to the chemical potential inside the swollen sample. This means, there must be an additional contribution to the Gibbs free energy of mixing (as is the case with the osmotic equilibrium) besides the common terms caused by mixing the (virtually) infinite-molar-mass polymer and the solvent. This additional part follows from the elastic deformation of the network. The different aspects of chemical and physical networks will not be discussed here, for some details please see Refs.201-205 The following text is restricted to the aspect of solvent activities only. One method to obtain solvent activities in swollen polymer networks in equilibrium is to apply vapor pressure measurements. This is discussed in detail above in the Subchapter 4.4.3.1.1 and most methods can be used also for network systems, especially all sorption methods, and need no further explanation. The VPO-technique can be applied for this purpose, e.g., Arndt.206,207 IGC-measurements are possible, too, if one realizes a definitely crosslinked polymer in the column, e.g., Refs.208-210 Besides vapor sorption/pressure measurements, definite swelling and/or deswelling experiments lead to information on solvent activities. Swelling experiments work with pure solvents, deswelling experiments use dilute solutions of macromolecules (which must not diffuse into or adsorb at the surface of the network) and allow measurements in dependence on concentration. Deswelling experiments can be made in dialysis cells to prevent diffusion into the network. The determination of the equilibrium swelling/deswelling concentration can be made by weighing, but, in most cases, by measuring the swelling degree. Some methods for measuring the swelling degree are: measuring the buoyancy difference of the sample in two different liquids of known density, e.g. Rijke and Prins,211 measuring the volume change by cathetometer, e.g., Schwarz et al.,212 measuring the volume change by electrical (inductive) measurements, e.g., Teitelbaum and Garwin.213 The swelling degree can be defined as the ratio of the masses (mass based degree) or of the volumes (volume based degree) of the swollen to the dry polymer network sample: Q m = 1 + m1 / m N

or

Qv = 1 + ν1 / ν n = 1 + (Q m − 1) ρ n / ρ 1

[4.4.62]

194

Christian Wohlfarth

where: Qm Qv m1 mN ν1 νN ρ1 ρN

mass based degree of swelling volume based degree of swelling absorbed mass of the solvent mass of the dry polymer network absorbed volume of the solvent volume of the dry polymer network density of the solvent density of the dry polymer network

Since Qv = 1/ϕ 2 , usually volume changes were measured. If the sample swells isotropically, the volume change can be measured by the length change in one dimension. Calculation of solvent activities from measurements of the swelling degree needs a statistical thermodynamic model. According to Flory,46 a closed thermodynamic cycle can be constructed to calculate the Gibbs free energy of swelling from the differences between the swelling process, the solution process of the linear macromolecules, the elastic deformation and the crosslinking. The resulting equation can be understood in analogy to the Flory-Huggins relation, Equation [4.4.13a] with r → ∞, and reads:

(

∆µ 1 / RT = ln(1 − ϕ 2 ) + ϕ 2 + χϕ 22 + ν cV1 Aηϕ12/ 3 − Bϕ 2

)

[4.4.63]

where: νc V1 η A B Mc ϕ2 χ

network density νc = ρN/Mc molar volume of the pure liquid solvent 1 at temperature T memory term microstructure factor volume factor molar mass of a network chain between two network knots equilibrium swelling concentration = 1/Qv Flory-Huggins χ-function

A numerical calculation needs knowledge of the solvent activity of the corresponding homopolymer solution at the same equilibrium concentration ϕ 2 (here characterized by the value of the Flory-Huggins χ-function) and the assumption of a deformation model that provides values of the factors A and B. There is an extensive literature for statistical thermodynamic models which provide, for example, Flory:46 A = 1 and B = 0.5; Hermans:214 A = 1 and B = 1; James and Guth215 or Edwards and Freed:216 A = 0.5 and B = 0. A detailed explanation was given recently by Heinrich et al.203 The swelling equilibrium depends on temperature and pressure. Both are related to the corresponding dependencies of solvent activity via its corresponding derivative of the chemical potential:  ∂T  1  ∂µ 1  T  ∂µ 1    =    =   ∂ϕ1  P ∆S1  ∂ϕ1  P ,T ∆H1  ∂ϕ1  P ,T

[4.4.64]

where: ϕ1 ∆S1 ∆H1

equilibrium swelling concentration of the solvent differential entropy of dilution at equilibrium swelling differential enthalpy of dilution at equilibrium swelling where ∆H1 = T∆S1

The first derivative in Equation [4.4.64] describes the slope of the swelling curve. Since the derivative of the chemical potential is always positive for stable gels, the positive


195

or negative slope is determined by the differential enthalpy or entropy of dilution, Rehage.217  ∂P  1   = V1 − V1  ∂ϕ1 T

(

)

 ∂µ 1     ∂ϕ1  P ,T

[4.4.65]

where: V1 V1

partial molar volume of the solvent in the polymer solution at temperature T molar volume of the pure solvent at temperature T

In analogy to membrane osmometry, swelling pressure measurements can be made to obtain solvent activities. Pure solvents as well as dilute polymer solutions can be applied. In the case of solutions, the used macromolecules must not diffuse into or adsorb at the surface of the network. Two types of swelling pressure apparatuses have been developed. The anisochoric swelling pressure device measures swelling degrees in dependence on pressure and swelling volume, e.g., Borchard.218 The isochoric swelling pressure device applies a compensation technique where the volume is kept constant by an external pressure which is measured, e.g., Borchard.219 Swelling pressures can also be measured by sedimentation equilibrium using an ultracentrifuge for details, please see Borchard.220,221 The swelling pressure πswell is directly related to the solvent activity by: ∆µ 1 = RT ln a1 = −V1 πswell

[4.4.66]

and may be in the range of some MPa. In comparison with all methods of determination of solvent activities from swelling equilibrium of network polymers, the gravimetric vapor sorption/pressure measurement is the easiest applicable procedure, gives the most reliable data, and should be preferred. 4.4.4 THERMODYNAMIC MODELS FOR THE CALCULATION OF SOLVENT ACTIVITIES OF POLYMER SOLUTIONS Since measurements of solvent activities of polymer solutions are very time-consuming and can hardly be made to cover the whole temperature and pressure ranges, good thermodynamic theories, and models are necessary, which are able to calculate them with sufficient accuracy and which can interpolate between and even extrapolate from some measured basic data over the complete phase and state ranges of interest for a special application. Many attempts have been made to find theoretical explanations for the non-ideal behavior of polymer solutions. There exist books and reviews on this topic, e.g., Refs.2,41,42,46-49 Therefore, only a short summary of some of the most important thermodynamic approaches and models will be given here. The following explanations are restricted to concentrated polymer solutions only because one has to describe mainly concentrated polymer solutions when solvent activities have to be calculated. For dilute polymer solutions, with the second virial coefficient region, Yamakawa’s book222 provides a good survey. There are two different approaches for the calculation of solvent activities of polymer solutions: (i) the approach which uses activity coefficients, starting from Equation [4.4.11] (ii) the approach which uses fugacity coefficients, starting from Equations [4.4.5 and 4.4.6].

196

Christian Wohlfarth

From the historical point of view and also from the number of applications in the literature, the common method is to use activity coefficients for the liquid phase, i.e., the polymer solution, and a separate equation-of-state for the solvent vapor phase, in many cases the truncated virial equation of state as for the data reduction of experimental measurements explained above. To this group of theories and models also free-volume models and lattice-fluid models will be added in this paper because they are usually applied within this approach. The approach where fugacity coefficients are calculated from one equation of state for both phases was applied to polymer solutions more recently, but it is the more promising method if one has to extrapolate over larger temperature and pressure ranges. Theories and models are presented below without going into details and without claiming completeness, since this text is not dedicated to theoretical problems but will only provide some help to calculate solvent activities. 4.4.4.1 Models for residual chemical potential and activity coefficient in the liquid phase Since polymer solutions in principle do not fulfill the rules of the ideal mixture but show strong negative deviations from Raoult’s law due to the difference in molecular size, the athermal Flory-Huggins mixture is usually applied as the reference mixture within polymer solution thermodynamics. Starting from Equation [4.4.11] or from  ∂n∆ mix G   RT ln a1 = RT ln x 1 γ 1 = ∆µ 1 = µ 1 − µ 10 =   ∂n1 T , P , nj≠1

[4.4.67]

where: a1 x1 γ1 µ1 µ 10 R T n1 n ∆ mixG

activity of the solvent mole fraction of the solvent activity coefficient of the solvent in the liquid phase with activity a1 = x1γ 1 chemical potential of the solvent chemical potential of the solvent at standard state conditions gas constant absolute temperature amount of substance (moles) of the solvent total amount of substance (moles) in the polymer solution molar Gibbs free energy of mixing,

the classical Flory-Huggins theory46,47 leads, for a truly athermal binary polymer solution, to:  1 ln a1athermal = ∆µ 1athermal / RT = ln(1 − ϕ 2 ) +  1 − ϕ 2  r

[4.4.68a]

  1   1 ln γ 1athermal = ln1 −  1 −  ϕ 2  +  1 −  ϕ 2   r   r

[4.4.68b]

or

where: ϕ2

volume fraction of the polymer


197

which is also called the combinatorial contribution to solvent activity or chemical potential, arising from the different configurations assumed by polymer and solvent molecules in solution, ignoring energetic interactions between molecules and excess volume effects. The Flory-Huggins derivation of the athermal combinatorial contribution contains the implicit assumption that the r-mer chains placed on a lattice are perfectly flexible and that the flexibility of the chain is independent of the concentration and of the nature of the solvent. Generalized combinatorial arguments for molecules containing different kinds of energetic contact points and shapes were developed by Barker223 and Tompa,224 respectively. Lichtenthaler et al.225 have used the generalized combinatorial arguments of Tompa to analyze VLE of polymer solutions. Other modifications have been presented by Guggenheim226,227 or Staverman228 (see below). The various combinatorial models are compared in a review by Sayegh and Vera.229 Recently, Freed and coworkers230-232 developed a lattice-field theory, that, in principle, provides an exact mathematical solution of the combinatorial Flory-Huggins problem. Although the simple Flory-Huggins expression does not always give the (presumably) correct, quantitative combinatorial entropy of mixing, it qualitatively describes many features of athermal polymer solutions. Therefore, for simplicity, it is used most in the further presentation of models for polymer solutions as reference state. The total solvent activity/activity coefficient/chemical potential is simply the sum of the athermal part as given above, plus a residual contribution: ln a1 = ln a1athermal + ln a1residual

[4.4.69a]

µ 1 = µ 10 + µ 1athermal + µ 1residual

[4.4.69b]

or

The residual part has to be explained by an additional model and a number of suitable models is now listed in the following text. The Flory-Huggins interaction function of the solvent is the residual function used first and is given by Equations [4.4.12 and 4.4.13] with µ 1residual / RT = χϕ 22 . It was originally based on van Laar’s concept of solutions where excess entropy and excess volume of mixing could be neglected and χ is represented only in terms of an interchange energy ∆ε/kT. In this case, the interchange energy refers not to the exchange of solvent and solute molecules but rather to the exchange of solvent molecules and polymer segments. For athermal solutions, χ is zero, and for mixtures of components that are chemically similar, χ is small compared to unity. However, χ is not only a function of temperature (and pressure) as was evident from this foundation, but it is also a function of composition and polymer molecular mass, see e.g., Refs.5,7,8 If we neglect these dependencies, then the Scatchard-Hildebrand theory,233,234 i.e., their solubility parameter concept, could be applied: µ 1residual / RT = (V1 / RT )(δ1 − δ 2 ) ϕ 22 2

[4.4.70]

with

(

δ1 = ∆vap U10 / V1

)

1/ 2

[4.4.71]

198

Christian Wohlfarth

where: V1 ∆ vap U10 δ1 δ2

molar volume of the pure liquid solvent 1 at temperature T molar energy of vaporization of the pure liquid solvent 1 at temperature T solubility parameter of the pure liquid solvent solubility parameter of the polymer

Solubility parameters of polymers cannot be calculated from energy of vaporization since polymers do not evaporate. Usually they have been measured according to Equation [4.4.70], but details need not be explained here. Equation [4.4.70] is not useful for an accurate quantitative description of polymer solutions but it provides a good guide for a qualitative consideration of polymer solubility. For good solubility, the difference between both solubility parameters should be small (the complete residual chemical potential term cannot be negative, which is one of the disadvantages of the solubility parameter approach). Several approximate generalizations have been suggested by different authors - a summary of all these models and many data can be found in the books by Barton.11,12 Calculations applying additive group contributions to obtain solubility parameters, especially of polymers, are also explained in the book by Van Krevelen.235 Better-founded lattice models have been developed in the literature. The ideas of Koningsveld and Kleintjens, e.g., Ref.,51 lead to useful and easy to handle expressions, as is given above with Equations [4.4.15, 4.4.17 and 4.4.46] that have been widely used, but mainly for liquid-liquid demixing and not so much for vapor-liquid equilibrium and solvent activity data. Comprehensive examples can be found in the books by Fujita41 or Kamide.42 The simple Flory-Huggins approach and the solubility parameter concept are inadequate when tested against experimental data for polymer solutions. Even for mixtures of n-alkanes, the excess thermodynamic properties cannot be described satisfactorily - Flory et al.236-239 In particular, changes of volume upon mixing are excluded and observed excess entropies of mixing often deviate from simple combinatorial contributions. To account for these effects, the PVT-behavior has to be included in theoretical considerations by an equation of state. Pure fluids have different free volumes, i.e., different degrees of thermal expansion depending on temperature and pressure. When liquids with different free volumes are mixed, that difference contributes to the excess functions. Differences in free volumes must be taken into account, especially for mixtures of liquids whose molecules differ greatly in size, i.e., the free volume dissimilarity is significant for polymer solutions and has important effects on their properties, such as solvent activities, excess volume, excess enthalpy and excess entropy. Additionally, the free volume effect is the main reason for liquid-liquid demixing with LCST behavior at high temperatures.240,241 Today, there are two principal ways to develop an equation of state for polymer solutions: first, to start with an expression for the canonical partition function utilizing concepts similar to those used by van der Waals (e.g., Prigogine,242 Flory et al.,236-239 Patterson,243,244 Simha and Somcynsky,245 Sanchez and Lacombe,246-248 Dee and Walsh,249 Donohue and Prausnitz,250 Chien et al.251), and second, which is more sophisticated, to use statistical thermodynamics perturbation theory for freely-jointed tangent-sphere chain-like fluids (e.g., Hall and coworkers,252-255 Chapman et al.,256-258 Song et al.259,260). A comprehensive review about equations of state for molten polymers and polymer solutions was given by Lambert et al. 261 Here, only some resulting equations will be summarized under the aspect of calculating solvent activities in polymer solutions. The theories that are usually applied within activity coefficient models are given now, the other theories are summarized in Subchapter 4.4.4.2.


199

The first successful theoretical approach of an equation of state model for polymer solutions was the Prigogine-Flory-Patterson theory.236-242 It became popular in the version by Flory, Orwoll and Vrij236 and is a van-der-Waals-like theory based on the corresponding-states principle. Details of its derivation can be found in numerous papers and books and need not be repeated here. The equation of state is usually expressed in reduced form and reads: ~ 1/ 3

~ ~

PV ~

T

V

=

~ 1/ 3

V

−

1

[4.4.72]

~ ~

− 1 VT

where the reduced PVT-variables are defined by ~

~

~

P = P / P *, T = T / T *, V = V / T *, P * V * = rcRT *

[4.4.73]

and where a parameter c is used (per segment) such that 3rc is the number of effective external degrees of freedom per (macro)molecule. This effective number follows from Prigogine’s approximation that external rotational and vibrational degrees of freedom can be considered as equivalent translational degrees of freedom. The equation of state is valid for liquid-like densities and qualitatively incorrect at low densities because it does not fulfill the ideal gas limit. To use the equation of state, one must know the reducing or characteristic parameters P*, V*, T*. These have to be fitted to experimental PVT-data. Parameter tables can be found in the literature - here we refer to the book by Prausnitz et al.,49 a review by Rodgers,262 and the contribution by Cho and Sanchez263 to the new edition of the Polymer Handbook. To extend the Flory-Orwoll-Vrij model to mixtures, one has to use two assumptions: (i) the hard-core volumes υ* of the segments of all components are additive and (ii) the intermolecular energy depends in a simple way on the surface areas of contact between solvent molecules and/or polymer segments. Without any derivation, the final result for the residual solvent activity in a binary polymer solution reads: residual 1

ln a

+

~ 1/ 3   P1*V1*  ~ V 1 − 1  1 1  = 3T 1 ln 1/ 3 + ~ −~ + ~   RT  V − 1  V 1 V  

V1*  Χ12 RT  V~

 2 PV * ~ ~  θ + 1 (V − V 1 )  2 RT 

[4.4.74]

where: X12 θ2

interaction parameter surface fraction of the polymer

The last term in Equation [4.4.74] is negligible at normal pressures. The reduced volume of the solvent 1 and the reduced volume of the mixture are to be calculated from the same equation of state, Equation [4.4.72], but for the mixture the following mixing rules have to be used (if random mixing is assumed):

200

Christian Wohlfarth

P * = P1* ψ 1 + P2* ψ 2 − ψ 1 θ 2 Χ12 T * = P* /

[(P ψ * 1

) (

[4.4.75a]

/ T1* + P2* ψ 2 / T2*

1

)]

[4.4.75b]

where the segment fractions ψi and the surface fractions θi have to be calculated according to:

∑n V / ∑ψ s

ψ i = n iVi* / θi = ψ is i

k

k

* k k

* = m iVspez ,i /

∑m V

* k spez , k

= x i ri /

∑x

r

k k

[4.4.76a] [4.4.76b]

where: mi xi ri si

mass of component i mole fraction of component i number of segments of component i, here with ri /rk = Vi* / Vk* and r1 = 1 number of contact sites per segment (proportional to the surface area per segment)

Now it becomes clear from Equation [4.4.74] that the classical Flory-Huggins χ-function (χψ 22 = ln a 1residual ) varies with composition, as found experimentally. However, to calculate solvent activities by applying this model, a number of parameters have to be considered. The characteristic parameters of the pure substances have to be obtained by fitting to experimental PVT-data as explained above. The number of contact sites per segment can be calculated from Bondi’s surface-to-volume parameter tables264 but can also be used as fitting parameter. The X12-interaction parameter has to be fitted to experimental data of the mixture. Fitting to solvent activities, e.g. Refs.,265,266 does not always give satisfactorily results. Fitting to data for the enthalpies of mixing gives comparable results.266 Fitting to excess volumes alone does not give acceptable results.142 Therefore, a modification of Equation [4.4.74] was made by Eichinger and Flory142 by appending the term -(V1* / R)Q12 θ 22 where the parameter Q12 represents the entropy of interaction between unlike segments and is an entropic contribution to the residual chemical potential of the solvent. By adjusting the parameter Q12, a better representation of solvent activities can be obtained. There are many papers in the literature that applied the Prigogine-Flory-Patterson theory to polymer solutions as well as to low-molecular mixtures. Various modifications and improvements were suggested by many authors. Sugamiya267 introduced polar terms by adding dipole-dipole interactions. Brandani268 discussed effects of non-random mixing on the calculation of solvent activities. Kammer et al.269 added a parameter reflecting differences in segment size. Shiomi et al.270,271 assumed non-additivity of the number of external degrees of freedom with respect to segment fraction for mixtures and assumed the sizes of hard-core segments in pure liquids and in solution to be different. Also Panayiotou272 accounted for differences in segment size by an additional parameter. Cheng and Bonner273 modified the concept to obtain an equation of state which provides the correct zero pressure limit of the ideal gas. An attractive feature of the theory is its straightforward extension to multi-component mixtures,274 requiring only parameters of pure components and binary ones as explained above. A general limitation is its relatively poor description of the compressibility behavior of polymer melts, as well as its deficiencies regarding the description of the pressure dependence of thermodynamic data of mixtures. Dee and Walsh249 developed a modified version of Prigogine’s cell model that provides an excellent description of the PVT-behavior of polymer melts:


~ 1/ 3

~ ~

PV ~

T

V

=

~ 1/ 3

V

−

− 0.8909q

201

  2  12045 . 1011 .  − ~  ~4  ~2 T V V 

[4.4.77]

where the reduced variables and characteristic parameters have the same definitions as in the Flory model above. Equation [4.4.77] is formally identical with Prigogine’s result, except for the additional constant parameter q, which can also be viewed as a correction to the hard-core cell volume. The value of q = 1.07 corresponds approximately to a 25% increase in the hard-core volume in comparison with the original Prigogine model. Characteristic parameters for this model are given in Refs.249,262 The final result for the residual solvent activity in a binary polymer solution reads: residual 1

ln a

+

V1* θ 22 RT

~ 1 /3    0.5055 0.5055   12045 P1*V1*  ~ V1 − 0.8909q  12045 . .   − + − = 3 T 1 ln 1/3 − ~ ~2 ~4  ~2    ~4 RT  V   V1 V V − 0.8909q  V1  

 * ~  12045 0.5055  PV1  .   Χ12 − TQ12 V   ~ 2 − ~ 4  + RT    V  V

~ ~  V −V 1   

[4.4.78]

The last term in Equation [4.4.78] is again negligible at normal pressures, which is the case for the calculation of solvent activities of common polymer solutions. The reduced volume of the mixture is to be calculated from the equation of state where the same mixing rules are valid, as given by Equations [4.4.75, 4.4.76] if random mixing is assumed. Equation [4.4.78] is somewhat more flexible than Equation [4.4.74]. Again, entropic parameter Q12 and interaction parameter X12 have to be fitted to experimental data of the mixture. There is not much experience with the model regarding thermodynamic data of polymer solutions because it was mainly applied to polymer blends, where it provides much better results than the simple Flory model. To improve on the cell model, two other classes of models were developed, namely, lattice-fluid and lattice-hole theories. In these theories, vacant cells or holes are introduced into the lattice to describe the extra entropy change in the system as a function of volume and temperature. The lattice size, or cell volume, is fixed so that the changes in volume can only occur by the appearance of new holes, or vacant sites, on the lattice. The most popular theories of such kind were developed by Simha and Somcynsky245 or Sanchez and Lacombe.246-248 The Sanchez-Lacombe lattice-fluid equation of state reads: ~ ~

~

~ 1 V− 1 1 = − 1 − V ln ~ − ~ ~ ~ r T V VT

PV

[4.4.79]

where the reduced parameters are given in Equation [4.4.73], but no c-parameter is included, and the size parameter, r, and the characteristic parameters are related by P * V * = (r / M ) RT *

[4.4.80]

202

Christian Wohlfarth

where: r M

size parameter (segment number) molar mass

In comparison with Equation [4.4.72], the size parameter remains explicit in the reduced equation of state. Thus, a simple corresponding-states principle is not, in general, satisfied. But, in principle, this equation of state is suitable for describing thermodynamic properties of fluids over an extended range of external conditions from the ordinary liquid to the gaseous state (it gives the correct ideal gas limit) and also to conditions above the critical point where the fluid is supercritical. Equation of state parameters for many liquids and liquid/molten polymers have recently been reported by Sanchez and Panayiotou275 and for polymers by Rodgers262 and by Cho and Sanchez.263 To extend the lattice fluid theory to mixtures, appropriate mixing rules are needed. There is a fundamental difficulty here, because the segment size of any component is not necessarily equal to that of another but the molecular hard-core volume of a component must not change upon mixing. Consequently, the segment number of a component in the mixture, ri, may differ from that for the pure fluid, ri0 . But, following the arguments given by Panayiotou,276 the number of segments may remain constant in the pure state and in the mixture. This assumption leads to a simpler formalism without worsening the quantitative character of the model. Thus, the following mixing rules may be applied: P * = P1* ψ 1 + P2* ψ 2 − ψ 1 ψ 2 Χ12

[4.4.81a]

V * = ΣΣψ i ψ jVij*

[4.4.81b]

1/ r = Σψ i / ri

[4.4.81c]

where Vii* = Vi* and Vij* provides an additional binary fitting parameter and Equation [4.4.80] provides the mixing rule for T*. The final result for the residual solvent activity in a binary polymer solution reads: residual 1

ln a

~ ~  ~  V ~   V 1 − 1  ~   V − 1 = r1 V − 1 ln ~  − r1 V 1 − 1 ln ~  − ln ~     V1  V   V1 

Χ r 1 1 − ~1  ~ − ~  + r1  ~12     V T 1 V V1 

~ ~ ~   2 r1 P 1 V −V 1  ψ + ~  2  T1

[4.4.82]

The last term in Equation (4.4.82) is again negligible at normal pressures. Various other approximations were given in the literature. For example, one can assume random mixing of contact sites rather than random mixing of segments,277,278 as well as non-random mixing.277,279 The model is applicable to solutions of small molecules as well as to polymer solutions. Like the Prigogine-Flory-Patterson equation of state, the lattice-fluid model and its variations have been used to correlate the composition dependence of the residual solvent activity.277,279 These studies show that again entropic parameter Q12 and interaction parameter X12 have to be fitted to experimental data of the mixture to provide better agreement


203

with measured solvent activities. The model and its modifications have been successfully used to represent thermodynamic excess properties, VLE and LLE for a variety of mixtures, as summarized by Sanchez and Panayiotou.275 In addition to mixtures of polymers with normal solvents, the model can also be applied to polymer-gas systems, Sanchez and Rodgers.280 In lattice-hole theories, vacant cells or holes are introduced into the lattice, which describe the major part of thermal expansion, but changes in cell volume are also allowed which influence excess thermodynamic properties as well. The hole-theory for polymeric liquids as developed by Simha and Somcynsky245 provides a very accurate equation of state that works much better than the Prigogine-Flory-Patterson equation of state or the Sanchez-Lacombe lattice-fluid model with respect to the precision how experimental PVT-data can be reproduced. However, the Dee-Walsh equation of state, Equation [4.4.77], with its more simple structure, works equally well. The Simha-Somcynsky equation of state must be solved simultaneously with an expression that minimizes the partition function with respect to the fraction of occupied sites and the final resulting equations for the chemical potential are more complicated. Details of the model will not be provided here. Characteristic parameters for many polymers have recently been given by Rodgers262 or Cho and Sanchez.263 The model is applicable to solutions of small molecules as well as to polymer solutions. Binary parameters have to be fitted to experimental data as with the models explained above. Again, one can assume random mixing of contact sites rather than random mixing of segments as well as non-random mixing, as was discussed, for example, by Nies and Stroeks281 or Xie et al.282,283 Whereas the models given above can be used to correlate solvent activities in polymer solutions, attempts also have been made in the literature to develop concepts to predict solvent activities. Based on the success of the UNIFAC concept for low-molecular liquid mixtures,284 Oishi and Prausnitz285 developed an analogous concept by combining the UNIFAC-model with the free-volume model of Flory, Orwoll and Vrij.236 The mass fraction based activity coefficient of a solvent in a polymer solution is given by: ln Ω 1 ln(a1 / w1 ) = ln Ω 1comb + ln Ω 1res + ln Ω 1fv

[4.4.83]

where: Ω1 Ω1comb Ω1res Ω1fv

mass fraction based activity coefficient of solvent 1 at temperature T combinatorial contribution to the activity coefficient residual contribution to the activity coefficient free-volume contribution to the activity coefficient

Instead of the Flory-Huggins combinatorial contribution, Equation [4.4.68], the Staverman relation228 is used. ln Ω 1comb = ln

ψ1 z θ ψM + q 1 ln 1 + l 1 − 1 1 w1 2 ψ1 w1

∑ j

ψ jM j wj

[4.4.84]

where the segment fractions ψi and the surface area fractions θi have to be calculated according to ψ i = (w i ri / M i ) / Σ(w k rk / M k )

[4.4.85a]

204

Christian Wohlfarth

θ i = (w i q i / M i ) / Σ(w k q k / M k )

[4.4.85b]

and the li-parameter is given by l i = (z / 2)(ri − q i ) − (ri − 1)

[4.4.85c]

where: i,j z qi Mi wi ri

components in the solution lattice coordination number (usually = 10) surface area of component i based on Bondi’s van-der-Waals surfaces molar mass of component i (for polymers the number average is recommended) mass fraction of component i segment number of component i based on Bondi’s van-der-Waals volumes

The molecules must be divided into groups as defined by the UNIFAC method. The segment surface areas and the segment volumes are calculated from Bondi’s tables264 according to ri = ∑ ν (κi ) Rk

and

k

q i = ∑ ν (ki ) Q k

[4.4.86]

k

where: k νk Rk Qk

number of groups of type k in the solution number of groups of type k in the molecule i Van-der-Waals group volume parameter for group k Van-der-Waals group surface parameter for group k

The residual activity coefficient contribution for each component is given by

[

ln Ω res = ∑ ν k( i ) ln Γk − ln Γk( i ) i k

]

[4.4.87]

where: Γk Γk(i)

the residual activity coefficient of group k in the defined solution the residual activity coefficient of group k in a reference solution containing pure component i only

The residual activity coefficient of group k in the given solution is defined by     a  ln Γk = Q k 1 − ln ∑ Λ m exp  − mk  T  m  

   Λ exp − a km   m    T   [4.4.88]   − ∑    a pm    m  − Λ exp    p ∑  T    p

where: amn and anm group interaction parameter pair between groups m and n group surface area fraction for group m Λm

The group surface area fraction Λm for group m in the solution is given by Λm =

Qm X m ∑ Qp X p p

[4.4.89a]


205

where: Xm

group mole fraction for group m

The group mole fraction Xm for group m in the solution is given by Xm =

∑ν j

(j) m

∑ ∑ν j

p

wj / Mj

(j) p

wj / Mj

[4.4.89b]

The residual activity coefficient of group k in reference solutions containing only component i, Γk(i) is similarly determined using Equations [4.4.88, 4.4.89], with the exception that the summation indices k, m, p refer only to the groups present in the pure component and the summations over each component j are calculated only for the single component present in the reference solution. The group interaction parameter pairs amn and anm result from the interaction between the groups m and n. These parameter are unsymmetric values that have to be fitted to experimental VLE-data of low-molecular mixtures. They can be taken from UNIFAC tables, e.g., Refs.2,284,286-289 and, additionally, they may be treated as temperature functions. The free-volume contribution, which is essential for nonpolar polymer solutions, follows, in principle, from Equation [4.4.74] with parameter X12 = 0 as applied by Raetzsch and Glindemann,290 or in a modified form from Equation [4.4.90] as introduced by Oishi and Prausnitz and used also in Danner’s Handbook.2

ln Q1fv

    ~ ~  ~ 1/ 3   V 1 −V   V 1 − 1 = 3c 1 ln  1/ 3  − c 1   ~ 1/ 3  ~  V~ − 1    V 1 −V 1          

[4.4.90]

where: c1

external degree of freedom parameter of the solvent 1, usually fixed = 1.1

To get a predictive model, the reduced volumes and the external degree of freedom parameter are not calculated from Flory’s equation of state, Equation [4.4.72], but from some simple approximations as given by the following relations:2 ~

Vi =

~

V=

υspez, i M i

[4.4.91a]

0.01942ri

∑υ w 0.01942∑ r w i

spez, i

i

i

i

i

/ Mi

[4.4.91b]

where: υspez,i ri Mi

specific volume of component i in m3/kg segment number of component i based on Bondi’s van-der-Waals volumes molar mass of component i (for polymers the number average is recommended)

206

Christian Wohlfarth

wi

mass fraction of component i

There has been a broad application of this group-contribution UNIFAC-fv concept to polymer solutions in the literature. Raetzsch and Glindemann290 recommended the use of the real free-volume relation from the Flory-Orwoll-Vrij model to account for realistic PVT-data. Problems arise for mixtures composed from chemically different components that posses the same groups, e.g., mixtures with different isomers. Kikic et al.291 discussed the influence of the combinatorial part on results obtained with UNIFAC-fv calculations. Gottlieb and Herskowitz292,293 gave some polemic about the special use of the c1-parameter within UNIFAC-fv calculations. Iwai et al.294,295 demonstrated the possible use of UNIFAC-fv for the calculation of gas solubilities and Henry’s constants using a somewhat different free-volume expression. Patwardhan and Belfiore296 found quantitative discrepancies for some polymer solutions. In a number of cases UNIFAC-fv predicted the occurrence of a demixing region during the calculation of solvent activities where experimentally only a homogeneous solution exists. Price and Ashworth297 found that the predicted variation of residual solvent activity with polymer molecular mass at high polymer concentrations is opposite to that measured for polydimethylsiloxane solutions. But, qualitative correct predictions were obtained for poly(ethylene glycol) solutions with varying polymer molecular mass.298-300 However, UNIFAC-fv is not capable of representing thermodynamic data of strongly associating or solvating systems. Many attempts have been made to improve the UNIFAC-fv model which cannot be listed here. A comprehensive review was given by Fried et al.301 An innovative method to combine the free-volume contribution within a corrected Flory-Huggins combinatorial entropy and the UNIFAC concept was found by Elbro et al.302 and improved by Kontogeorgis et al.303 These authors take into account the free-volume dissimilarity by assuming different van der Waals hard-core volumes (again from Bondi’s tables264) for the solvent and the polymer segments

(

/3 Vi fv = q i Vi1/ 3 − Vi1,vdW

ϕ fvi = x iVi fv /

∑x V j

j

)

3

[4.4.92a]

fv j

[4.4.92b]

where: qi xi Vi Vi,vdW

surface area of component i based on Bondi’s van der Waals surfaces mole fraction of component i molar volume of component i van der Waals hard-core molar volume of component i

and introduced these free-volume terms into Equation [4.4.68] to obtain a free-volume corrected Flory-Huggins combinatorial term:

(

) + 1− (ϕ

(

)

ln γ fvi = ln ϕ fvi / x i

fv i

/ xi

)

[4.4.93a]

or

(

ln Ω fvi = ln ϕ fvi / w i + 1 − ϕ fvi / x i

)

[4.4.93b]


207

To obtain the complete activity coefficient, only the residual term from the UNIFAC model, Equation [4.4.87], has to be added. An attempt to incorporate differences in shape between solvent molecules and polymer segments was made by Kontogeorgis et al.302 by adding the Staverman correction term to obtain: ln γ fvi = ln

ϕ fvi ϕ fv zq + 1− i − i 2 xi xi

 ψi ψ  ln + 1− i θi  θi

  

[4.4.93c]

where the segment fractions ψi and the surface area fractions θi have to be calculated according to Equations [4.4.85a+b]. Using this correction, they get somewhat better results only when Equation [4.4.93] leads to predictions lower than the experimental data. Different approaches utilizing group contribution methods to predict solvent activities in polymer solutions have been developed after the success of the UNIFAC-fv model. Misovich et al.304 have applied the Analytical Solution of Groups (ASOG) model to polymer solutions. Recent improvements of polymer-ASOG have been reported by Tochigi et al.305-307 Various other group-contribution methods including an equation-of-state were developed by Holten-Anderson et al.,308,309 Chen et al.,310 High and Danner,311-313 Tochigi et al.,314 Lee and Danner,315 Bertucco and Mio,316 or Wang et al.,317 respectively. Some of them were presented again in Danner’s Handbook.2 Detail are not provided here. 4.4.4.2 Fugacity coefficients from equations of state Total equation-of-state approaches usually apply equations for the fugacity coefficients instead of relations for chemical potentials to calculate thermodynamic equilibria and start from Equations [4.4.2 to 6]. Since the final relations for the fugacity coefficients are usually much more lengthy and depend, additionally, on the chosen mixing rules, only the equations of state are listed below. Fugacity coefficients have to be derived by solving Equation [4.4.6]. After obtaining the equilibrium fugacities of the liquid mixture at equilibrium temperature and pressure, the solvent activity can be calculated from Equation [4.4.1]. The standard state fugacity of the solvent can also be calculated from the same equation of state by solving the same equations but for the pure liquid. Details of this procedure can be found in textbooks, e.g., Refs.318,319 Equations of state for polymer systems that will be applied within such an approach have to be valid for the liquid as well as for the gaseous state like lattice-fluid models based on Sanchez-Lacombe theory, but not the free-volume equations based on Prigogine-Flory-Patterson theory, as stated above. However, most equations of state applied within such an approach have not been developed specially for polymer systems, but, first, for common non-electrolyte mixtures and gases. Today, one can distinguish between cubic and non-cubic equations of state for phase equilibrium calculations in polymer systems. Starting from the free-volume idea in polymeric systems, non-cubic equations of state should be applied to polymers. Thus, the following text presents first some examples of this class of equations of state. Cubic equations of state came later into consideration for polymer systems, mainly due to increasing demands from engineers and engineering software where three-volume-roots equations of state are easier to solve and more stable in computational cycles. About ten years after Flory’s development of an equation of state for polymer systems, one began to apply methods of thermodynamic perturbation theory to calculate the thermo-

208

Christian Wohlfarth

dynamic behavior of polymer systems. The main goal was first to overcome the restrictions of Flory’s equation of state to the liquid state, to improve the calculation of the compressibility behavior with increasing pressure and to enable calculations of fluid phase equilibria at any densities and pressures from the dilute gas phase to the compressed liquid including molecules differing considerably in size, shape, or strength of intermolecular potential energy. More recently, when more sophisticated methods of statistical mechanics were developed, deeper insights into liquid structure and compressibility behavior of model polymer chains in comparison to Monte Carlo modelling results could be obtained applying thermodynamic perturbation theory. Quite a lot of different equations of state have been developed up to now following this procedure; however, only a limited number was applied to real systems. Therefore, only some summary and a phenomenological presentation of some equations of state which have been applied to real polymer fluids should be given here, following their historical order. The perturbed-hard-chain (PHC) theory developed by Prausnitz and coworkers in the late 1970s320-322 was the first successful application of thermodynamic perturbation theory to polymer systems. Since Wertheim’s perturbation theory of polymerization323 was formulated about 10 years later, PHC theory combines results from hard-sphere equations of simple liquids with the concept of density-dependent external degrees of freedom in the Prigogine-Flory-Patterson model for taking into account the chain character of real polymeric fluids. For the hard-sphere reference equation the result derived by Carnahan and Starling324 was applied, as this expression is a good approximation for low-molecular hard-sphere fluids. For the attractive perturbation term, a modified Alder’s325 fourth-order perturbation result for square-well fluids was chosen. Its constants were refitted to the thermodynamic equilibrium data of pure methane. The final equation of state reads: mA 4y 2 − 2y PV = 1+ c + c ∑ ∑ m nmn 3 ~ ~ RT n m (y − 1) V T

[4.4.94]

where: y

c V0 Anm

packing fraction with y = V/(V0τ) and τ = (π / 6 )2 0 .5 = 0.7405 (please note that in a number of original papers in the literature the definition of y within this kind of equations is made by the reciprocal value, i.e., τV0/V) degree of freedom parameter, related to one chain-molecule (not to one segment) hard-sphere volume for closest packing empirical coefficients from the attractive perturbation term ~

The reduced volume is again defined by V = V/V0 and the reduced temperature by ~ T = T / T*. The coefficients Anm are given in the original papers by Beret320,321 and are considered to be universal constants that do not depend on the chemical nature of any special substance. The remaining three characteristic parameters, c, T* and V0 , have to be adjusted to experimental PVT-data of the polymers or to vapor-liquid equilibrium data of the pure solvents. Instead of fitting the c-parameter, one can also introduce a parameter P* by the relation P* = cRT*/V0. In comparison with Flory’s free-volume equation of state, PHC-equation of state is additionally applicable to gas and vapor phases. It fulfills the ideal gas limit, and it describes the PVT-behavior at higher pressures better and without the need of temperature and/or pressure-dependent characteristic parameters, such as with Flory’s model. Values for characteristic parameters of polymers and solvents can be found in the original literature. A review for the PHC-model was given by Donohue and Vimalchand,326 where a


209

number of extensions and applications also are summarized. Application to mixtures and solutions needs mixing rules for the characteristic parameters and introduction of binary fitting parameters322,327,328 (details are not given here). Examples for applying PHC to polymer solutions are given by Liu and Prausnitz328 or Iwai, Arai and coworkers.329-331 The chain-of-rotators (COR) equation of state was developed by Chao and coworkers332 as an improvement of the PHC theory. It introduces the non-spherical shape of molecules into the hard-body reference term and describes the chain molecule as a chain of rotators with the aim of an improved model for calculating fluid phase equilibria, PVT and derived thermodynamic properties, at first only for low-molecular substances. Instead of hard spheres, the COR-model uses hard dumbbells as reference fluid by combining the result of Boublik and Nezbeda333 with the Carnahan-Starling equation for a separate consideration of rotational degrees of freedom; however, still in the sense of Prigogine-Flory-Patterson regarding the chain-character of the molecules. It neglects the effect of rotational motions on intermolecular attractions; however, the attractive portion of the final equation of state has an empirical dependence on rotational degrees of freedom given by the prefactor of the double sum. For the attractive perturbation term, a modified Alder’s fourth-order perturbation result for square-well fluids was chosen, additionally improved by an empirical temperature-function for the rotational part. The final COR equation reads: 2 4y 2 − 2y PV  α − 1 3y + 3αy − (α + 1) = 1+ +c  3 3 RT  2  (y − 1) (y − 1) ~ ~ mA  c  + 1 +  B 0 + B1 / T + B 2 T   ∑ ∑ m nmn 2  n m ~ ~  V T

[4.4.95]

where: y

c V0 Anm B0,B1,B2 α

packing fraction with y = V/(V0τ) and τ = (π / 6 )2 0 .5 = 0.7405 (please note that in a number of original papers in the literature the definition of y within this kind of equations is made by its reciprocal value, i.e., τV0/V) degree of freedom parameter, related to one chain-molecule (not to one segment) hard-sphere volume for closest packing empirical coefficients from the attractive perturbation term empirical coefficients for the temperature dependence of the rotational part accounts for the deviations of the dumbbell geometry from a sphere

As can be seen from the structure of the COR equation of state, the Carnahan-Starling term becomes very small with increasing chain length, i.e., with increasing c, and the rotational part is the dominant hard-body term for polymers. The value of c is here a measure of rotational degrees of freedom within the chain (and related to one chain-molecule and not to one segment). It is different from the meaning of the c-value in the PHC equation. Its exact value is not known a priori as chain molecules have a flexible structure. The value of α for the various rotational modes is likewise not precisely known. Since α and c occur together in the product c(α - 1), departure of real rotators from a fixed value of α is compensated for by the c-parameter after any fitting procedure. As usual, the value of α is assigned a constant value of 1.078 calculated according to the dumbbell for ethane as representative for the rotating segments of a hydrocarbon chain. The coefficients Anm and the three parameters B0,

210

Christian Wohlfarth

B1, B2 were refitted to the thermodynamic equilibrium data of pure methane, ethane, and propane.332 Both Anm matrices for PHC and COR equation of state contain different numerical values. The remaining three characteristic parameters, c, T* and V0, have to be adjusted to experimental equilibrium data. Instead of fitting the c-parameter, one can also introduce a parameter P* by the relation P* = cRT*/V0. Characteristic parameters for many solvents and gases are given by Chien et al.332 or Masuoka and Chao.334 Characteristic parameters of more than 100 polymers and many solvents are given by Wohlfarth and coworkers,335-348 who introduced segment-molar mixing rules and group-contribution interaction parameters into the model and applied it extensively to polymer solutions at ordinary pressures as well as at high temperatures und pressures, including gas solubility and supercritical solutions. They found that it may be necessary sometimes to refit the pure-component characteristic data of a polymer to some VLE-data of a binary polymer solution to calculate correct solvent activities, because otherwise demixing was calculated. Refitting is even more necessary when high-pressure fluid phase equilibria have to be calculated using this model. A group-contribution COR equation of state was developed Pults et al.349,350 and extended into a polymer COR equation of state by Sy-Siong-Kiao et al.351 This equation of state is somewhat simplified by replacing the attractive perturbation term by the corresponding part of the Redlich-Kwong equation of state. 2 a(T ) 4y 2 − 2y PV  α − 1 3y + 3αy − (α + 1) [4.4.96] = 1+ +c −  3 3 RT RT [V + b (T )]  2  (y − 1) (y − 1)

where: a b c y α

attractive van der Waals-like parameter excluded volume van der Waals-like parameter degree of freedom parameter, related to one chain-molecule (not to one segment) packing fraction accounts for the deviations of the dumbbell geometry from a sphere

Exponential temperature functions for the excluded volume parameter b and the attractive parameter a were introduced by Novenario et al.352-354 to apply this equation of state also to polar and associating fluids. Introducing a group-contribution concept leads to segment-molar values of all parameters a, b, c which can easily be fitted to specific volumes of polymers.351,354 The statistical associating fluid theory (SAFT) is the first and the most popular approach that uses real hard-chain reference fluids, including chain-bonding contributions. Its basic ideas have been developed by Chapman et al.256-258 Without going into details, the final SAFT equation of state is constructed from four terms: a segment term that accounts for the non-ideality of the reference term of non-bonded chain segments/monomers as in the equations shown above, a chain term that accounts for covalent bonding, and an association term that accounts for hydrogen bonding. There may be an additional term that accounts for other polarity effects. A dispersion term is also added that accounts for the perturbing potential, as in the equations above. A comprehensive summary is given in Praunsitz’s book.49 Today, there are different working equations based on the SAFT approach. Their main differences stem from the way the segment and chain terms are estimated. The most common version is the one developed by Huang and Radosz,355 applying the fourth-order perturbation approach as in COR or PHC above, but with new refitted parameters to argon, as given


211

by Chen and Kreglewski,356 and a hard-sphere pair-correlation function for the chain term as following the arguments of Wertheim. The Huang-Radosz-form of the SAFT-equation of state without an association term reads:355 n

mDnm  u  4y 2 − 2y 5y − 2 PV [4.4.97] = 1+ r + (1 − r ) + r∑ ∑ 3 RT (y − 1)(2y − 1) n m V~ m kT  (y − 1) where: Dnm k r

empirical coefficients from the attractive perturbation term Boltzmann’s constant chain segment number

In comparison to the PHC equation of state, the new term between the Carnahan-Starling term and the double sum accounts for chain-bonding. The terms are proportional to the segment number, r , of the chain molecule. However, the hard-sphere volume, V0, is now a slight function of temperature which is calculated according to the result of Chen and Kreglewski:356 V0 = V 00 [1 − 012 . exp(−3u 0 / kT )]

3

[4.4.98]

where: V00 u0

hard-sphere volume at T = 0 K well-depth of a square-well potential u/k = u0/k (1 + 10K/T) with 10K being an average for all chain molecules.

The ratio u0/k or u/k is analogous to the characteristic parameter T* in the equations above. There are two additional volume and energy parameters if association is taken into account. In its essence, the SAFT equation of state needs three pure component parameters which have to be fitted to equilibrium data: V00, u0/k and r. Fitting of the segment number looks somewhat curious to a polymer scientist, but it is simply a model parameter, like the c-parameter in the equations above, which is also proportional to r. One may note additionally that fitting to specific volume PVT-data leads to a characteristic ratio r/M (which is a specific r-value), as in the equations above, with a specific c-parameter. Several modifications and approximations within the SAFT-framework have been developed in the literature. Banaszak et al.357-359 or Shukla and Chapman360 extended the concept to copolymers. Adidharma and Radosz361 provides an engineering form for such a copolymer SAFT approach. SAFT has successfully applied to correlate thermodynamic properties and phase behavior of pure liquid polymers and polymer solutions, including gas solubility and supercritical solutions by Radosz and coworkers355,357-359,361-368 Sadowski et al.369 applied SAFT to calculate solvent activities of polycarbonate solutions in various solvents and found that it may be necessary to refit the pure-component characteristic data of the polymer to some VLE-data of one binary polymer solution to calculate correct solvent activities, because otherwise demixing was calculated. Groß and Sadowski370 developed a “Perturbed-Chain SAFT” equation of state to improve for the chain behavior within the reference term to get better calculation results for the PVT- and VLE-behavior of polymer systems. McHugh and coworkers applied SAFT extensively to calculate the phase behavior of polymers in supercritical fluids, a comprehensive summary is given in the review by Kirby and McHugh.371 They also state that characteristic SAFT parameters for polymers from PVT-data lead to

212

Christian Wohlfarth

wrong phase equilibrium calculations and, therefore, also to wrong solvent activities from such calculations. Some ways to overcome this situation and to obtain reliable parameters for phase equilibrium calculations are provided in Ref.,371 together with examples from the literature that will not be repeated here. The perturbed-hard-sphere-chain (PHSC) equation of state is a hard-sphere-chain theory that is somewhat different to SAFT. It is based on a hard-sphere chain reference system and a van der Waals-type perturbation term using a temperature-dependent attractive parameter a(T) and a temperature-dependent co-volume parameter b(T). Song et al.259,260 applied it to polymer systems and extended the theory also to fluids consisting of heteronuclear hard chain molecules. The final equation for pure liquids or polymers as derived by Song et al. is constructed from three parts: the first term stems (as in PHC, COR or SAFT) from the Carnahan-Starling hard-sphere monomer fluid, the second is the term due to covalent chain-bonding of the hard-sphere reference chain and the third is a van der Waals-like attraction term (more details are given also in Prausnitz’s book49):  (1 − η / 2)  r 2 a(T ) 4η − 2η 2 PV 1 = 1+ r + − r − 1 − ( ) 3 RT (1 − η) 3  RTV  (1 − η)

[4.4.99]

where: η a r

reduced density or packing fraction attractive van der Waals-like parameter chain segment number

The reduced density or packing fraction η is related to an effective and temperature-dependent co-volume b(T) by η = r b(T)ρ/4, with ρ being the number density, i.e., the number of molecules per volume. However, PHSC-theory does not use an analytical intermolecular potential to estimate the temperature dependence of a(T) and b(T). Instead, empirical temperature functions are fitted to experimental data of argon and methane (see also49). We note that the PHSC equation of state is again an equation where three parameters have to be fitted to thermodynamic properties: σ, ε/k and r. These may be transformed into macroscopic reducing parameters for the equation of state by the common relations T*=ε/k, P* = 3ε/2πσ3 and V* = 2πrσ3/3. Parameter tables are given in Refs.86,260,372-374 PHSC was successfully applied to calculate solvent activities in polymer solutions, Gupta and Prausnitz.86 Lambert et al.374 found that it is necessary to adjust the characteristic parameters of the polymers when liquid-liquid equilibria should correctly be calculated. Even with simple cubic equations of state, a quantitative representation of solvent activities for real polymer solutions can be achieved, as was shown by Tassios and coworkers.375,376 Using generalized van der Waals theory, Sako et al.377 obtained a three-parameter cubic equation of state which was the first applied to polymer solutions: a(T ) PV V − b + bc = − RT V −b RT (V + b ) where: a b

attractive van der Waals-like parameter excluded volume van der Waals-like parameter

[4.4.100]


c

213

3c is the total number of external degrees of freedom per molecule

When c = 1, Equation [4.4.100] reduces to the common Soave-Redlich-Kwong (SRK) equation of state.378 Temperature functions and combining/mixing rules for parameters a,b,c are not discussed here because quite different approximations may be used. Problems, how to fit these parameters to experimental PVT-data for polymers, have been discussed by several authors.375-380 Orbey and Sandler380 applied the Peng-Robinson equation of state as modified by Stryjek and Vera381 (PRSV): a(T ) PV V = − 2 RT V − b RT V + 2bV − b 2

(

)

[4.4.101]

to calculate solvent activities in polymer solutions using Wong-Sandler mixing rules382 that combine the equation of state with excess energy models (EOS/GE-mixing rules). They have shown that a two-parameter version can correlate the solvent partial pressure of various polymer solutions with good accuracy over a range of temperatures and pressures with temperature-independent parameters. Harrismiadis et al.379 worked out some similarities between activity coefficients derived from van der Waals like equations-of-state and Equations (4.4.92 and 93), i.e., the Elbro-fv model. Zhong and Masuoka383 combined SRK equation of state with EOS/GE-mixing rules and the UNIFAC model to calculate Henry’s constants of solvents and gases in polymers. Additionally, they developed new mixing rules for van der Waals-type two-parameter equations of state (PRSV and SRK) which are particularly suitable for highly asymmetric systems, i.e., also polymer solutions, and demonstrated that only one adjustable temperature-independent parameter is necessary for calculations within a wide range of temperatures.384 In a following paper,385 some further modifications and improvements could be found. Orbey et al.386 successfully proposed some empirical relations for PRSV-equation-of-state parameters with polymer molar mass and specific volume to avoid any special parameter fitting for polymers and introduced a NRTL-like local-composition term into the excess energy part of the mixing rules for taking into account of strong interactions, for example, in water + poly(propylene glycol)s. They found infinite-dilution activity coefficient data, i.e., Henry’s constants, to be most suitable for fitting the necessary model parameter.386 Orbey et al.387 summarized three basic conclusions for the application of cubic equations of state to polymer solutions: (i) These models developed for conventional mixtures can be extended to quantitatively describe VLE of polymer solutions if carefully selected parameters are used for the pure polymer. On the other hand, pure-component parameters of many solvents are already available and VLE between them is well represented by these cubic equations of state. (ii) EOS/GE-mixing rules represent an accurate way of describing phase equilibria. Activity coefficient expressions are more successful when they are used in this format than directly in the conventional gamma-phi approach. (iii) It is not justifiable to use multi-parameter models, but it is better to limit the number of parameters to the number of physically meaningful boundary conditions and calculate them according to the relations dictated by these boundary conditions.

214

Christian Wohlfarth

4.4.4.3 Comparison and conclusions The simple Flory-Huggins χ-function, combined with the solubility parameter approach may be used for a first rough guess about solvent activities of polymer solutions, if no experimental data are available. Nothing more should be expected. This also holds true for any calculations with the UNIFAC-fv or other group-contribution models. For a quantitative representation of solvent activities of polymer solutions, more sophisticated models have to be applied. The choice of a dedicated model, however, may depend, even today, on the nature of the polymer-solvent system and its physical properties (polar or non-polar, association or donor-acceptor interactions, subcritical or supercritical solvents, etc.), on the ranges of temperature, pressure and concentration one is interested in, on the question whether a special solution, special mixture, special application is to be handled or a more universal application is to be found or a software tool is to be developed, on numerical simplicity or, on the other hand, on numerical stability and physically meaningful roots of the non-linear equation systems to be solved. Finally, it may depend on the experience of the user (and sometimes it still seems to be a matter of taste). There are deficiencies in all of these theories given above. These theories fail to account for long-range correlations between polymer segments which are important in dilute solutions. They are valid for simple linear chains and do not account for effects like chain branching, rings, dentritic polymers. But, most seriously, all of these theories are of the mean-field type that fail to account for the contributions of fluctuations in density and composition. Therefore, when these theories are used in the critical region, poor results are often obtained. Usually, critical pressures are overestimated within VLE-calculations. Two other conceptually different mean-field approximations are invoked during the development of these theories. To derive the combinatorial entropic part correlations between segments of one chain that are not nearest neighbors are neglected (again, mean-field approximations are therefore not good for a dilute polymer solution) and, second, chain connectivity and correlation between segments are improperly ignored when calculating the potential energy, the attractive term. Equation-of-state approaches are preferred concepts for a quantitative representation of polymer solution properties. They are able to correlate experimental VLE data over wide ranges of pressure and temperature and allow for physically meaningful extrapolation of experimental data into unmeasured regions of interest for application. Based on the experience of the author about the application of the COR equation-of-state model to many polymer-solvent systems, it is possible, for example, to measure some vapor pressures at temperatures between 50 and 100oC and concentrations between 50 and 80 wt% polymer by isopiestic sorption together with some infinite dilution data (limiting activity coefficients, Henry’s constants) at temperatures between 100 and 200oC by IGC and then to calculate the complete vapor-liquid equilibrium region between room temperature and about 350oC, pressures between 0.1 mbar and 10 bar, and solvent concentration between the common polymer solution of about 75-95 wt% solvent and the ppm-region where the final solvent and/or monomer devolatilization process takes place. Equivalent results can be obtained with any other comparable equation of state model like PHC, SAFT, PHSC, etc. The quality of all model calculations with respect to solvent activities depends essentially on the careful determination and selection of the parameters of the pure solvents, and also of the pure polymers. Pure solvent parameter must allow for the quantitative calculation of pure solvent vapor pressures and molar volumes, especially when equation-of-state


215

approaches are used. Pure polymer parameters strongly influence the calculation of gas solubilities, Henry’s constants, and limiting solvent activities at infinite dilution of the solvent in the liquid/molten polymer. Additionally, the polymer parameters mainly determine the occurrence of a demixing region in such model calculations. Generally, the quantitative representation of liquid-liquid equilibria is a much more stringent test for any model, what was not discussed here. To calculate such equilibria it is often necessary to use some mixture properties to obtain pure-component polymer parameters. This is necessary because, at present, no single theory is able to describe correctly the properties of a polymer in both the pure molten state and in the highly dilute solution state. Therefore, characteristic polymer parameters from PVT-data of the melt are not always meaningful for the dilute polymer solution. Additionally, characteristic polymer parameters from PVT-data also may lead to wrong results for concentrated polymer solutions because phase equilibrium calculations are much more sensitive to variations in pure component parameters than polymer densities. All models need some binary interaction parameters that have to be adjusted to some thermodynamic equilibrium properties since these parameters are a priori not known (we will not discuss results from Monte Carlo simulations here). Binary parameters obtained from data of dilute polymer solutions as second virial coefficients are often different from those obtained from concentrated solutions. Distinguishing between intramolecular and intermolecular segment-segment interactions is not as important in concentrated solutions as it is in dilute solutions. Attempts to introduce local-composition and non-random-mixing approaches have been made for all the theories given above with more or less success. At least, they introduce additional parameters. More parameters may cause a higher flexibility of the model equations but leads often to physically senseless parameters that cause troubles when extrapolations may be necessary. Group-contribution concepts for binary interaction parameters in equation of state models can help to correlate parameter sets and also data of solutions within homologous series. 4.4.5 REFERENCES 1

2 3 4 5 6 7 8 9 10 11

Wen Hao, H. S. Elbro, P. Alessi, Polymer Solution Data Collection, Pt.1: Vapor-Liquid Equilibrium, Pt.2: Solvent Activity Coefficients at Infinite Dilution, Pt. 3: Liquid-Liquid Equilibrium, DECHEMA Chemistry Data Series, Vol. XIV, Pts. 1, 2+3, DECHEMA, Frankfurt/M., 1992. R. P. Danner, M. S. High, Handbook of Polymer Solution Thermodynamics, DIPPR, AIChE, New York, 1993. Ch. Wohlfarth, Vapour-liquid equilibrium data of binary polymer solutions, Physical Science Data 44, Elsevier, Amsterdam, 1994. J. Brandrup, E. H. Immergut, E. A. Grulke (eds.), Polymer Handbook, 4th ed., J. Wiley & Sons., Inc., New York, 1999. N. Schuld, B. A. Wolf, in Polymer Handbook, J. Brandrup, E. H. Immergut, E. A. Grulke (eds.), 4th ed., J. Wiley & Sons., Inc., New York, 1999, pp. VII/247-264. R. A. Orwoll, in Polymer Handbook, J. Brandrup, E. H. Immergut, E. A. Grulke (eds.), 4th ed., J. Wiley & Sons., Inc., New York, 1999, pp. VII/649-670. R. A. Orwoll, P. A. Arnold, in Physical Properties of Polymers Handbook, J. E. Mark (Ed.), AIP Press, Woodbury, New York, 1996, pp. 177-198. R. A. Orwoll, Rubber Chem. Technol., 50, 451 (1977). M. D. Lechner, E. Nordmeier, D. G. Steinmeier, in Polymer Handbook, J. Brandrup, E. H. Immergut, E. A. Grulke (eds.), 4th ed., J. Wiley & Sons., Inc., New York, 1999, pp. VII/85-213. D. C. Bonner, J. Macromol. Sci. - Revs. Macromol. Chem., C, 13, 263 (1975). A. F. M. Barton, CRC Handbook of Polymer-Liquid Interaction Parameters and Solubility Parameters, CRC Press, Boca Raton, 1990.

216

12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52

Christian Wohlfarth

A. F. M. Barton, CRC Handbook of Solubility Parameters and Other Cohesion Parameters, 2nd Ed., CRC Press, Boca Raton, 1991. J.-M. Braun, J. E. Guillet, Adv. Polym. Sci., 21, 108 (1976). A. E. Nesterov, J. S. Lipatov, Obrashchennaya Gasovaya Khromatografiya v Termodinamike Polimerov, Naukova Dumka, Kiev, 1976. D. C. Locke, Adv. Chromatogr., 14, 87 (1976). D. G. Gray, Progr. Polym. Sci., 5, 1 (1977). J. E. G. Lipson, J. E. Guillet, Developm. Polym. Character., 3, 33 (1982). J. S. Aspler, Chromatogr. Sci., 29, 399 (1985). A. E. Nesterov, Obrashchennaya Gasovaya Khromatografiya Polimerov, Naukova Dumka, Kiev, 1988. R. Vilcu, M. Leca, Polymer Thermodynamics by Gas Chromatography, Elsevier, Amsterdam, 1989. G. J. Price, Adv. Chromatogr., 28, 113 (1989). D. R. Lloyd, T. C. Ward, H. P. Schreiber, C. C. Pizana (eds.), Inverse Gas Chromatography, ACS Symp. Ser., 391, 1989. P. Munk, in Modern Methods of Polymer Characterization, H. G. Barth, J. W. Mays (eds.), J. Wiley & Sons, Inc., New York, 1991, pp.151-200. Bincai Li, Rubber Chem. Technol., 69, 347 (1996). Z. Y. Al-Saigh, Int. J. Polym. Anal. Charact. 3, 249 (1997). R. A. Pethrick, J. V. Dawkins (Eds.), Modern Techniques for Polymer Characterization, J. Wiley & Sons, Inc., Chichester, 1999. G. D. Wignall, in Encyclopedia of Polymer Science and Engineering, 2nd Ed., J. Kroschwitz (Ed.), J. Wiley & Sons, Inc., New York, 1987, Vol. 10, pp. 112-184. G. C. Berry, in Encyclopedia of Polymer Science and Engineering, 2nd Ed., J. Kroschwitz (Ed.), J. Wiley & Sons, Inc., New York, 1987, Vol. 8, pp. 721-794. M. B. Huglin (Ed.), Light Scattering from Polymer Solutions, Academic Press, Inc., New York, 1972. E. F. Casassa, G. C. Berry, in Polymer Molecular Weights, Pt. 1, Marcel Dekker, Inc., New York, 1975, pp. 161-286. P. Kratochvil, Classical Light Scattering from Polymer Solutions, Elsevier, Amsterdam, 1987. B. Chu, Laser Light Scattering, Academic Press, Inc., New York, 1991. A. R. Cooper, in Encyclopedia of Polymer Science and Engineering, 2nd Ed., J. Kroschwitz (Ed.), J. Wiley & Sons, Inc., New York, 1987, Vol. 10, pp. 1-19. C. A. Glover, in Polymer Molecular Weights, Pt. 1, Marcel Dekker, Inc., New York, 1975, pp. 79-159. H. Fujita, Foundations of Ultracentrifugal Analysis, J.Wiley & Sons, Inc., New York, 1975. S. E. Harding, A. J. Rowe, J. C. Horton, Analytical Ultracentrifugation in Biochemistry and Polymer Science, Royal Society of Chemistry, Cambridge, 1992. P. Munk, in Modern Methods of Polymer Characterization, H. G. Barth, J. W. Mays (eds.), J. Wiley & Sons, Inc., New York, 1991, pp. 271-312. E.T.Adams, in Encyclopedia of Polymer Science and Engineering, 2nd Ed., J. Kroschwitz (Ed.), J. Wiley & Sons, Inc., New York, 1987, Vol. 10, pp. 636-652. M. P. Tombs, A. R. Peacock, The Osmotic Pressure of Macromolecules, Oxford University Press, London, 1974. J. W. Mays, N. Hadjichristidis, in Modern Methods of Polymer Characterization, H. G. Barth, J. W. Mays (eds.), J.Wiley & Sons, Inc., New York, 1991, pp. 201-226. H. Fujita, Polymer Solutions, Elsevier, Amsterdam, 1990. K. Kamide, Thermodynamics of Polymer Solutions, Elsevier, Amsterdam, 1990 Y. Einaga, Progr. Polym. Sci., 19, 1 (1994). R. Koningsveld, Adv. Colloid Interfacial Sci., 2, 151 (1968). B. A. Wolf, Adv. Polym. Sci., 10, 109 (1972). P. J. Flory, Principles of Polymer Chemistry, Cornell University Press, Ithaca, 1953. M. L. Huggins, Physical Chemistry of High Polymers, J.Wiley & Sons., Inc., New York, 1958. H. Tompa, Polymer Solutions, Butterworth, London, 1956. J. M. Prausnitz, R. N. Lichtenthaler, E. G. de Azevedo, Molecular Thermodynamics of Fluid Phase Equilibria, 3rd Ed., Prentice Hall, Upper Saddle River, New Jersey, 1999. R. M. Masegosa, M. G. Prolongo, A. Horta, Macromolecules, 19, 1478 (1986). R. Koningsveld, L. A. Kleintjens, Macromolecules, 4, 637 (1971). M. T. Raetzsch, H. Kehlen, in Encyclopedia of Polymer Science and Engineering, 2nd Ed., J. Kroschwitz (Ed.), J. Wiley & Sons, Inc., New York, 1989, Vol. 15, pp. 481-491.


53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108

217

M. T. Raetzsch, H. Kehlen, Progr. Polym. Sci., 14, 1 (1989). M. T. Raetzsch, Ch. Wohlfarth, Adv. Polym. Sci., 98, 49 (1990). A. G. Williamson, in Experimental Thermodynamics, Vol.II, Experimental Thermodynamics of non-reacting Fluids, B. Le Neindre, B. Vodar (eds.), Butterworth, London, pp. 749-786. K. Schmoll, E. Jenckel, Ber. Bunsenges. Phys. Chem., 60, 756 (1956). K. Schotsch, B. A. Wolf, Makromol. Chem., 185, 2161 (1984). E. Killmann, F. Cordt, F. Moeller, Makromol. Chem., 191, 2929 (1990). J. H. Baxendale, E. V. Enustun, J. Stern, Phil. Trans. Roy. Soc. London, A, 243, 169 (1951). M. L. McGlashan, A. G. Williamson, Trans. Faraday Soc., 57, 588 (1961). M. L. McGlashan, K. W. Morcom, A. G. Williamson, Trans. Faraday Soc., 57, 581, (1961). H. Takenaka, J. Polym. Sci., 24, 321 (1957). R. J. Kokes, A. R. Dipietro, F. A. Long, J. Amer. Chem. Soc., 75, 6319 (1953). P. W. Allen, D. H. Everett, M. F. Penney, Proc. Roy. Soc. London A, 212, 149 (1952). D. H. Everett, M. F. Penney, Proc. Roy. Soc. London A, 212, 164 (1952). C. A. Haynes, R. A. Beynon, R. S. King, H. W. Blanch, J. M. Praunsitz, J. Phys. Chem., 93, 5612 (1989). W. R. Krigbaum, D. O. Geymer, J. Amer. Chem. Soc., 81, 1859 (1959). M. L. Lakhanpal, B. E. Conway, J. Polym. Sci., 46, 75 (1960). K. Ueberreiter, W. Bruns, Ber. Bunsenges. Phys. Chemie, 68, 541 (1964). R. S. Jessup, J. Res. Natl. Bur. Stand., 60, 47 (1958). Van Tam Bui, J. Leonard, Polym. J., 21, 185 (1989). C. E. H. Bawn, R. F. J. Freeman, A. R. Kamaliddin, Trans. Faraday Soc., 46, 677 (1950). G. Gee, L. R. J. Treloar, Trans. Faraday Soc., 38, 147 (1942). C. Booth, G. Gee, G. Holden, G. R. Williamson, Polymer, 5, 343 (1964). G. N. Malcolm, C. E. Baird, G. R. Bruce, K. G. Cheyne, R. W. Kershaw, M. C. Pratt, J. Polym. Sci., Pt. A-2, 7, 1495 (1969). L.R. Ochs, M. Kabri-Badr, H. Cabezas, AIChE-J., 36, 1908 (1990). C. Grossmann, R. Tintinger, G. Maurer, Fluid Phase Equil., 106, 111 (1995). Z. Adamcova, Sci. Pap. Prag. Inst. Chem. Technol., N2, 63 (1976). Z. Adamcova, Adv. Chem., 155, 361 (1976). K. Kubo, K. Ogino, Polymer, 16, 629 (1975). C. G. Panayiotou, J. H. Vera, Polym. J., 16, 89 (1984). M. T. Raetzsch, G. Illig, Ch. Wohlfarth, Acta Polymerica, 33, 89 (1982). P. J. T. Tait, A. M. Abushihada, Polymer, 18, 810 (1977). J. H. Van der Waals, J. J. Hermans, Rec. Trav. Chim. Pays-Bas, 69, 971 (1950). D. C. Bonner, J. M. Prausnitz, J. Polym. Sci., Polym. Phys. Ed., 12, 51 (1974). R. B. Gupta, J. M. Prausnitz, J. Chem. Eng. Data, 40, 784 (1995). S. Behme, G. Sadowski, W. Arlt, Chem.-Ing.-Techn., 67, 757 (1995). A. J. Ashworth, D. H. Everett, Trans. Faraday Soc., 56, 1609 (1960) J. S. Vrentas, J. L. Duda, S. T. Hsieh, Ind. Eng. Chem., Process Des. Dev., 22, 326 (1983). Y. Iwai, Y. Arai, J. Chem. Eng. Japan, 22, 155 (1989). J. W. McBain, A. M. Bakr, J. Amer. Chem. Soc., 48, 690 (1926). S. Hwang, J. Kim, K.-P. Yoo, J. Chem. Eng. Data, 43, 614 (1998). J. Kim, K. C. Joung, S. Hwang, W. Huh, C. S. Lee, K.-P. Yoo, Korean J. Chem. Eng., 15, 199 (1998). Y. C. Bae, J. J. Shim, D. S. Soane, J. M. Prausnitz, J. Appl. Polym. Sci., 47, 1193 (1993). A. J. Ashworth, J. Chem. Soc., Faraday Trans. I, 69, 459 (1973). A. J. Ashworth, G. J. Price, Thermochim. Acta, 82, 161 (1984). S. Saeki, J. C. Holste, D. C. Bonner, J. Polym. Sci., Polym. Phys. Ed., 20, 793, 805 (1982). S. Saeki, J. C. Holste, D. C. Bonner, J. Polym. Sci., Polym. Phys. Ed., 19, 307 (1981). S. Saeki, J. Holste, D. C. Bonner, J. Polym. Sci., Polym. Phys. Ed., 21, 2049 (1983) H. C. Wong, S. W. Campbell, V. R. Bhetnanabotla, Fluid Phase Equil., 139, 371 (1997). S. P. V. N. Mikkilineni, D.A. Tree, M. S. High, J. Chem. Eng. Data, 40, 750 (1995). O. Smidsrod, J. E. Guillet, Macromolecules, 2, 272 (1969). Z. Y. Al-Saigh, P. Munk, Macromolecules, 17, 803 (1984). P. Munk, P. Hattam, Q. Du, J. Appl. Polym. Sci., Appl. Polym. Symp., 43, 373 (1989). D. Glindemann, Dissertation, TH Leuna-Merseburg, 1979. N. F. Brockmeier, R. W. McCoy, J. A. Meyer, Macromolecules, 5, 130, 464 (1972). N. F. Brockmeier, R. E. Carlson, R. W. McCoy, AIChE-J., 19, 1133 (1973). J. R. Conder, J. R. Purnell, Trans. Faraday Soc., 64, 1505, 3100 (1968).

218

Christian Wohlfarth

109 110 111 112 113 114 115 116 117

J. R. Conder, J. R. Purnell, Trans. Faraday Soc., 65, 824, 839 (1969). G. J. Price, J. E. Guillet, J. Macromol. Sci.-Chem., A23, 1487 (1986). D. C. Bonner, N. F. Brockmeier, Ind. Eng. Chem., Process Des. Developm., 16, 180 (1977). S. Dincer, D. C. Bonner, R. A. Elefritz, Ind. Eng. Chem., Fundam., 18, 54 (1979). W. A. Ruff, C. J. Glover, A. T. Watson, AIChE-J., 32, 1948, 1954 (1986). L. L. Joffrion, C. J. Glover, Macromolecules, 19, 1710 (1986). T. K. Tsotsis, C. Turchi, C. J. Glover, Macromolecules, 20, 2445 (1987). R. P. Danner, F. Tihminlioglu, R. K. Surana, J. L. Duda, Fluid Phase Equil., 148, 171 (1998). J. R. Conder, C. L. Young, Physicochemical Measurements by Gas Chromatography, J.Wiley & Sons, Inc., New York, 1979. C. Barth, R. Horst, B. A. Wolf, J. Chem. Thermodyn., 30, 641 (1998). H.-M. Petri, N. Schuld, B. A. Wolf, Macromolecules, 28, 4975 (1995). B. Kolb, L. S. Ettre, Static Headspace Gas Chromatography - Theory and Practice, Wiley-VCH, Weinheim 1997. H. Wang, K. Wang, H. Liu, Y. Hu, J. East China Univ. Sci. Technol., 23, 614 (1997). T. Katayama, K. Matsumura, Y. Urahama, Kagaku Kogaku, 35, 1012 (1971). K. Matsumura, T. Katayama, Kagaku Kogaku, 38, 388 (1974). A. V. Hill, Proc. Roy. Soc. London, Ser. A, 127, 9 (1930). E. Regener, Ch. Wohlfarth, Wiss. Zeitschr. TH Leuna-Merseburg, 30, 211 (1988). J. Gaube, A. Pfennig, M. Stumpf, J.Chem. Eng. Data, 38, 163 (1993). J. Gaube, A. Pfennig, M. Stumpf, Fluid Phase Equil., 83, 365 (1993). A. Eliassi, H. Modarress, G. A. Mansoori, J. Chem. Eng. Data, 35, 52 (1999). V. Majer, V. Svoboda, Enthalpies of Vaporization of Organic Compounds, Blackwell Sci. Publ., Oxford, 1985. T. E. Daubert, R. P. Danner, Physical and Thermodynamic Properties of Pure Chemicals. Data Compilation, Hemisphere Publ.Corp., New York, 1989 and several Supplements up to 1999, data are online available from DIPPR database. B. H. Bersted, J. Appl. Polym. Sci., 17, 1415 (1973). B. H. Bersted, J. Appl. Polym. Sci., 18, 2399 (1974). R. V. Figini, Makromol. Chem., 181, 2049 (1980). R. V. Figini, M. Marx-Figini, Makromol. Chem., 182, 437 (1980). M. Marx-Figini, R. V. Figini, Makromol. Chem., 181, 2401 (1980). P. J. T. Tait, A. M. Abushihada, Polymer, 18, 810 (1977). A. J. Ashworth, C.-F. Chien, D. L. Furio, D. M. Hooker, M. M. Kopecni, R. J. Laub, G. J. Price, Macromolecules, 17, 1090 (1984). R. N. Lichtenthaler, D. D. Liu, J. M. Prausnitz, Macromolecules, 7, 565 (1974). I. Noda, N. Kato, T. Kitano, M. Nagasawa, Macromolecules, 14, 668 (1981). H. Vink, Europ. Polym. J., 7, 1411 (1971) and 10, 149 (1974). A. Nakajima, F. Hamada, K. Yasue, K. Fujisawa, T. Shiomi, Makromol. Chem., 175, 197 (1974). B. E. Eichinger, P. J. Flory, Trans. Faraday Soc., 64, 2035, 2053, 2061, 2066 (1968). H. Hoecker, P. J. Flory, Trans. Faraday Soc., 67, 2270 (1971). P. J. Flory, H. Daoust, J. Polym. Sci., 25, 429 (1957). B. H. Zimm, I. Myerson, J. Amer. Chem. Soc., 68, 911 (1946). W. H. Stockmayer, E. F. Casassa, J. Chem. Phys., 20, 1560 (1952). G. Strobl, The Physics of Polymers, Springer Vlg., Berlin, 1996. B. H. Zimm, J. Chem. Phys., 16, 1093 (1948). B. A. Wolf, H.-J. Adams, J. Chem. Phys., 75, 4121 (1981). M. Lechner, G. V. Schulz, Europ. Polym. J., 6, 945 (1970). Y. Nakamura, T. Norisuye, A. Teramoto, J. Polym. Sci., Polym. Phys. Ed., 29, 153 (1991). Th. G. Scholte, Eur. Polym. J., 6, 1063 (1970). H. Hasse, H.-P. Kany, R. Tintinger, G. Maurer, Macromolecules, 28, 3540 (1995). H.-P. Kany, H. Hasse, G. Maurer, J. Chem. Eng. Data, 35, 230 (1999). Th. G. Scholte, J. Polym. Sci., Pt. A-2, 9, 1553 (1971). Th. G. Scholte, J. Polym. Sci., Pt. C, 39, 281 (1972). K. W. Derham, J. Goldsborough, M. Gordon, Pure Appl. Chem., 38, 97 (1974). M. Gordon, P. Irvine, J. W. Kennedy, J. Polym. Sci., Pt. C, 61, 199 (1977). M. Gordon, NATO ASI Ser., Ser. E, 89, 429 (1985).

118 119 120 121 122 123 124 125 126 127 128 129 130

131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159


219

160 H. Galina, M. Gordon, B. W. Ready, L. A. Kleintjens, in Polymer Solutions, W. C. Forsman (Ed.), Plenum Press, New York, 1986, pp. 267-298. 161 A. Guinier, G. Fournet, Small Angle Scattering of X-rays, J.Wiley & Sons, Inc., New York, 1955. 162 S. H. Chen, S. Yip (eds.), Spectroscopy in Biology and Chemistry, Neutron, X-ray and Laser, Academic Press, New York, 1974. 163 O. Glatter, O. Kratky (eds.), Small-Angle X-ray Scattering, Academic Press, London, 1982. 164 R. G. Kirste, W. Wunderlich, Z. Phys. Chem. N.F., 58, 133 (1968). 165 R. G. Kirste, G. Wild, Makromol.Chem., 121, 174 (1969). 166 G. Meyerhoff, U. Moritz, R. G. Kirste, W. Heitz, Europ. Polym. J., 7, 933 (1971). 167 K. Okano, T. Ichimura, K. Kurita, E. Wada, Polymer, 28, 693 (1987). 168 T. Ichimura, K. Okano, K. Kurita, E. Wada, Polymer, 28, 1573 (1987). 169 M. Takada, K. Okano, K. Kurita, Polym. J., 26, 113 (1994). 170 F. Horkay, A.-M. Hecht, H. B. Stanley, E. Geissler, Eur. Polym. J., 30, 215 (1994). 171 J. S. Higgins, A. Macconachie, in Polymer Solutions, W. C. Forsman (Ed.), Plenum Press, New York, 1986, pp. 183-238. 172 B. Hammouda, Adv. Polym. Sci., 106, 87 (1993). 173 N. Vennemann, M. D. Lechner, R. C. Oberthuer, Polymer, 28, 1738 (1987). 174 A. M. Pedley, J. S. Higgins, D. G. Pfeiffer, A. R. Rennie, Macromolecules, 23, 2492 (1990). 175 B. J. Berne, R. Pecora, Dynamic Light Scattering, J.Wiley & Sons, Inc., New York, 1976. 176 Th. G. Scholte, J. Polym. Sci., Pt. A-2, 8, 841 (1970). 177 B. J. Rietveld, Th. G. Scholte, J. P. L. Pijpers, Brit. Polym. J., 4, 109 (1972). 178 H.-G. Elias, H. Lys, Makromol.Chem., 92, 1 (1966). 179 T. Hirao, A. Teramoto, T. Sato, T. Norisuye, T. Masuda, T. Higashimura, Polym. J., 23, 925 (1991). 180 Y. Doucet, in Experimental Thermodynamics, Vol.II, Experimental Thermodynamics of non-reacting Fluids, B. Le Neindre, B. Vodar (eds.), Butterworth, London, pp. 835-900. 181 T. Kawai, J. Polym. Sci., 32, 425 (1958). 182 Y. Hoei, Y. Ikeda, M. Sasaki, J. Phys. Chem. B, 103, 5353 (1999). 183 R. Koningsveld, in Polymer Science, E. D. Jenkins (Ed.), North-Holland Publ.Comp., Amsterdam, 1972, pp. 1047-1134. 184 A. Imre, W. A. Van Hook, J. Polym. Sci., Pt. B, Polym. Sci., 34, 751 (1996) and J. Phys. Chem. Ref. Data, 25, 637 (1996). 185 O. Pfohl, T. Hino, J. M. Prausnitz, Polymer, 36, 2065 (1995). 186 N. Kuwahara, D. V. Fenby, M. Tamsuy, B. Chu: J. Chem. Phys., 55, 1140 (1971). 187 R. Koningsveld, A. J. Staverman, J. Polym. Sci., Pt. A-2, 6, 349 (1968). 188 B. A. Wolf, M. C. Sezen, Macromolecules, 10, 1010 (1977). 189 G. F. Alfrey, W. G. Schneider, Discuss. Faraday Soc., 15, 218 (1953). 190 S. C. Greer, D. T. Jacobs, J. Phys. Chem., 84, 2888 (1980). 191 D. T. Jacobs, S. C. Greer, Phys. Rev. A, 24, 2075 (1981). 192 B. Muessig, H. Wochnowski, Angew. Makromol. Chem., 104, 203 (1982). 193 G. D. Rehage, D. Moeller, D. Ernst, Makromol. Chem., 88, 232 (1965). 194 Th. G. Scholte, R. Koningsveld, Kolloid Z. Z. Polym., 218, 58 (1968). 195 B. J. Rietfeld, Brit. Polym. J., 6, 181 (1974). 196 M. Gordon, L. A. Kleintjens, B. W. Ready, J. A. Torkington, Brit. Polym. J., 10, 170 (1978). 197 R. Koningsveld, L. A. Kleintjens, M. H. Onclin, J. Macromol. Sci. Phys. B, 18, 357 (1980). 198 P. G. de Gennes, Scaling Concepts in Polymer Physics, Cornell Univ. Press, Ithaca, 1979. 199 R. Koningsveld, A. J. Staverman, J. Polym. Sci., Pt. A-2, 6, 325 (1968). 200 B. A. Wolf, Makromol. Chem., 128, 284 (1969). 201 L. R. G. Treloar, The Physics of Rubber Elasticity, 3rd Ed., Clarendon Press, Oxford, 1975. 202 K. Dusek, W. Prins, Adv. Polym. Sci., 6, 1 (1969). 203 G. Heinrich, E. Straube, G. Helmis, Adv. Polym. Sci., 84, 33 (1988). 204 K. Dusek (Ed.), Responsive Gels I and II, Adv. Polym. Sci., 109 and 110 (1996). 205 G. Rehage, Ber. Bunsenges. Phys. Chem., 81, 969 (1977). 206 K. F. Arndt, J. Schreck, Acta Polymerica, 36, 56 (1985), 37, 500 (1986). 207 K. F. Arndt, I. Hahn, Acta Polymerica, 39, 560 (1988). 208 R. Sanetra, B. N. Kolarz, A. Wlochowicz, Polimery, 34, 490 (1989). 209 D. G. Gray, J. E. Guillet, Macromolecules, 5, 316 (1972). 210 M. Galin, Macromolecules, 10, 1239 (1977). 211 A. M. Rijke, W. Prins, J. Polym. Sci., 59, 171 (1962).

220

212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265

Christian Wohlfarth J. Schwarz, W. Borchard, G. Rehage, Kolloid-Z. Z. Polym., 244, 193 (1971). B. J. Teitelbaum, A. E. Garwin, Vysokomol. Soedin., Ser. A, 10, 1684 (1968). J. J. Hermans, J. Polym. Sci., 59, 191 (1962). H. M. James, E. Guth, J. Chem. Phys., 21, 1039 (1953). S. F. Edwards, E. K. Freed, J. Phys. C, 3, 739, 750 (1970). G. Rehage, Kolloid-Z. Z. Polym., 199, 1 (1964). W. Borchard, Dissertation, RWTH Aachen, 1966. W. Borchard, Progr. Colloid Polym. Sci., 57, 39 (1975). W. Borchard, Progr. Colloid Polym. Sci., 86, 84 (1991). W. Borchard, H. Coelfen, Makromol. Chem., Macromol. Symp., 61, 143 (1992). H. Yamakawa, Modern Theory of Polymer Solutions, Harper & Row, New York, 1971. J. A. Barker, J. Chem. Phys., 20, 1256 (1952). H. Tompa, Trans. Faraday Soc., 48, 363 (1952). R. N. Lichtenthaler, D. S. Abrams, J. M. Prausnitz, Can. J. Chem., 53, 3071 (1973). E. A. Guggenheim, Proc. Roy. Soc. London, A, 183, 203 (1944). E. A. Guggenheim, Mixtures, Clarendon Press, Oxford, 1952. A. J. Staverman, Recl. Trav. Chim. Pays-Bas, 69, 163 (1950). S. G. Sayegh, J. H. Vera, Chem. Eng. J., 19, 1 (1980). K. F. Freed, J. Phys. A, Math. Gen., 18, 871 (1985). M. G. Bawendi, K. F. Freed, J. Chem. Phys., 87, 5534 (1987), 88, 2741 (1988). W. G. Madden, A. I. Pesci, K. F. Freed, Macromolecules, 23, 1181 (1990). G. Scatchard, Chem. Rev., 44, 7 (1949). J. H. Hildebrand, R. L. Scott, The Solubility of Nonelectrolytes, 3rd Ed., Reinhold Publ. Corp., New York, 1950. D. W. Van Krevelen, Properties of Polymers, 3rd Ed., Elsevier, Amsterdam, 1990. P. J. Flory, R. A. Orwoll, A. Vrij, J. Amer. Chem. Soc., 86, 3507, 3515 (1964). P. J. Flory, J. Amer. Chem. Soc., 87, 1833 (1965). P. J. Flory, R. A. Orwoll, J. Amer. Chem. Soc., 89, 6814, 6822 (1964). P. J. Flory, Discuss. Faraday Soc., 49, 7 (1970). P. I. Freeman, J. S. Rowlinson, Polymer, 1, 20 (1960). A. H. Liddell, F. L. Swinton, Discuss. Faraday Soc., 49, 215 (1970). I. Prigogine, The Molecular Theory of Solutions, North-Holland Publ., Amsterdam, 1957. D. Patterson, Macromolecules, 2, 672 (1969). D. Patterson, G. Delmas, Discuss. Faraday Soc., 49, 98 (1970). R. Simha, T. Somcynsky, Macromolecules, 2, 342 (1969). I. C. Sanchez, R. H. Lacombe, J. Phys. Chem., 80, 2352, 2568 (1976). I. C. Sanchez, R. H. Lacombe, J. Polym. Sci., Polym. Lett. Ed., 15, 71 (1977). I. C. Sanchez, R. H. Lacombe, Macromolecules, 11, 1145 (1978). G. T. Dee, D. J. Walsh, Macromolecules, 21, 811, 815 (1988). M. D. Donohue, J. M. Prausnitz, AIChE-J., 24, 849 (1978). C. H. Chien, R. A. Greenkorn, K. C. Chao, AIChE-J. 29, 560 (1983). K. G. Honnell, C. K. Hall, J. Chem. Phys., 90, 1841 (1989). H. S. Gulati, J. M. Wichert, C. K. Hall, J. Chem. Phys., 104, 5220 (1996). H. S. Gulati, C. K. Hall, J. Chem. Phys., 108, 5220 (1998). A. Yethiraj, C. K. Hall, J. Chem. Phys., 95, 1999 (1991). W. G. Chapman, G. Jackson, K. E. Gubbins, Mol. Phys., 65, 1057 (1988). W. G. Chapman, K. E. Gubbins, G. Jackson, M. Radosz, Fluid Phase Equil., 52, 31 (1989). W. G. Chapman, K. E. Gubbins, G. Jackson, M. Radosz, Ind. Eng. Chem. Res., 29, 1709 (1990). Y. Song, S. M. Lambert, J. M. Prausnitz, Macromolecules, 27, 441 (1994). Y. Song, S. M. Lambert, J. M. Prausnitz, Ind. Eng. Chem. Res., 33, 1047 (1994). S. M. Lambert, Y. Song, J. M. Prausnitz, in Equations of State for Fluids and Fluid Mixtures, J. V. Sengers et al. (eds.), Elsevier, Amsterdam, 2000, in press. P. A. Rodgers, J. Appl. Polym. Sci., 48, 1061 (1993). J. Cho, I. C. Sanchez, in Polymer Handbook, J. Brandrup, E. H. Immergut, E. A. Grulke (eds.), 4th ed., J. Wiley & Sons., Inc., New York, 1999, pp. VI/591-601. A. Bondi, Physical Properties of Molecular Crystals, Liquids and Glasses, John Wiley & Sons, New York, 1968. D. C. Bonner, J. M. Prausnitz, AIChE-J., 19, 943 (1973).


266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319

221

M. T. Raetzsch, M. Opel, Ch. Wohlfarth, Acta Polymerica, 31, 217 (1980). K. Sugamiya, Makromol. Chem., 178, 565 (1977). U. Brandani, Macromolecules, 12, 883 (1979). H.-W. Kammer, T. Inoue, T. Ougizawa, Polymer, 30, 888 (1989). T. Shiomi, K. Fujisawa, F. Hamada, A. Nakajima, J. Chem. Soc. Faraday Trans II, 76, 895 (1980). F. Hamada, T. Shiomi, K. Fujisawa, A. Nakajima, Macromolecules, 13, 279 (1980). C. G. Panayiotou, J. Chem. Soc. Faraday Trans. II, 80, 1435 (1984). Y. L. Cheng, D. C. Bonner, J. Polym. Sci., Polym. Phys. Ed., 15, 593 (1977), 16, 319 (1978). J. Pouchly, D. Patterson, Macromolecules, 9, 574 (1976). I. C. Sanchez, C. G. Panayiotou, in Models for Thermodynamics and Phase Equilibrium Calculations, S. I. Sandler (Ed.), Marcel Dekker, New York, 1994, pp. 187-285. C. G. Panayiotou, Makromol. Chem., 187, 2867 (1986). C. G. Panayiotou, J. H. Vera, Polym. J., 14, 681 (1982). C. G. Panayiotou, J. H. Vera, Can. J. Chem. Eng., 59, 501 (1981). C. G. Panayiotou, Macromolecules, 20, 861 (1987). I. C. Sanchez, P. A. Rodgers, Pure Appl. Chem., 62, 2107 (1990). E. Nies, A. Stroeks, Macromolecules, 23, 4088, 4092 (1990). H. Xie, E. Nies, Macromolecules, 26, 1689 (1993). H. Xie, R. Simha, Polym. Int., 44, 348 (1997). A. Fredenslund, J. Gmehling, P. Rasmussen, Vapor-Liquid Equilibria Using UNIFAC, Elsevier Sci. Publ., New York, 1977. T. Oishi, J. M. Prausnitz, Ind. Eng. Chem. Process Des. Dev., 17, 333 (1978). S. Skjold-Jorgensen, B. Kolbe, J. Gmehling, P. Rasmussen, Ind. Eng. Chem. Process Des. Dev., 18, 714 (1979). J. Gmehling, P. Rasmussen, A. Fredenslund, Ind. Eng. Chem. Process Des. Dev., 21, 118 (1982). E. A. Macedo, U. Weidlich, J. Gmehling, P. Rasmussen, Ind. Eng. Chem. Process Des. Dev., 22, 676 (1983). D. Tiegs, J. Gmehling, P. Rasmussen, A. Fredenslund, Ind. Eng. Chem. Res., 26, 159 (1987). M. T. Raetzsch, D. Glindemann, Acta Polymerica, 30, 57 (1979). I. Kikic, P. Alessi, P.Rasmussen, A. Fredenslund, Can. J. Chem. Eng., 58, 253 (1980). M. Gottlieb, M. Herskowitz, Ind. Eng. Chem. Process Des. Dev., 21, 536 (1982). M. Gottlieb, M. Herskowitz, Macromolecules, 14, 1468 (1981). Y. Iwai, M. Ohzono, Y. Arai, Chem. Eng. Commun., 34, 225 (1985). Y. Iwai, Y. Arai, J. Chem. Eng. Japan, 22, 155 (1989). A. A. Patwardhan, L. A. Belfiore, J. Polym. Sci. Pt. B Polym. Phys., 24, 2473 (1986). G. J. Price, A. J. Ashworth, Polymer, 28, 2105 (1987). M. T. Raetzsch, D. Glindemann, E. Hamann, Acta Polymerica, 31, 377 (1980). E. L. Sorensen, W. Hao, P. Alessi, Fluid Phase Equil., 56, 249 (1990). M. Herskowitz, M. Gottlieb, J. Chem. Eng. Data, 30, 233 (1985). J. R. Fried, J. S. Jiang, E. Yeh, Comput. Polym. Sci., 2, 95 (1992). H. S. Elbro, A. Fredenslund, P. A. Rasmussen, Macromolecules, 23, 4707 (1990). G. M. Kontogeorgis, A. Fredenslund, D. P. Tassios, Ind. Eng. Chem. Res., 32, 362 (1993). M. J. Misovich, E. A. Grulke, R. F. Blanks, Ind. Eng. Chem. Process Des. Dev., 24, 1036 (1985). K. Tochigi, S. Kurita, M. Ohashi, K. Kojima, Kagaku Kogaku Ronbunshu, 23, 720 (1997). J. S. Choi, K. Tochigi, K. Kojima, Fluid Phase Equil., 111, 143 (1995). B. G. Choi, J. S. Choi, K. Tochigi, K. Kojima, J. Chem. Eng. Japan, 29, 217 (1996). J. Holten-Andersen, A. Fredenslund, P. Rasmussen, G. Carvoli, Fluid Phase Equil., 29, 357 (1986). J. Holten-Andersen, A. Fredenslund, P. Rasmussen, Ind. Eng. Chem. Res., 26, 1382 (1987). F. Chen, A. Fredenslund, P. Rasmussen, Ind. Eng. Chem. Res., 29, 875 (1990). M. S. High, R. P. Danner, Fluid Phase Equil., 53, 323 (1989). M. S. High, R. P. Danner, AIChE-J., 36, 1625 (1990). M. S. High, R. P. Danner, Fluid Phase Equil., 55, 1 (1990). K. Tochigi, K. Kojima, T. Sako, Fluid Phase Equil., 117, 55 (1996). B.-C. Lee, R. P. Danner, Fluid Phase Equil., 117, 33 (1996). A. Bertucco, C. Mio, Fluid Phase Equil., 117, 18 (1996). W. Wang, X. Liu, C. Zhong, C. H. Twu, J. E. Coon, Fluid Phase Equil., 144, 23 (1998). J. M. Smith, H. C. Van Ness, M. M. Abbott, Introduction to Chemical Engineering Thermodynamics, 5th Ed., McGraw-Hill, Singapore, 1996. S. M. Walas, Phase equilibria in chemical engineering, Butterworth, Boston, 1985.

222

320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377

Christian Wohlfarth S. Beret, J. M. Prausnitz, AIChE-J., 21, 1123 (1975). S. Beret, J. M. Prausnitz, Macromolecules, 8, 878 (1975). M. D. Donohue, J. M. Prausnitz, AIChE-J., 24, 849 (1978). M. S. Wertheim, J.Chem.Phys., 87, 7323 (1987). N. F. Carnahan, K. E. Starling, J.Chem.Phys., 51, 635 (1969). B. J. Alder, D. A. Young, M. A. Mark, J.Chem.Phys., 56, 3013 (1972). M. D. Donohue, P. Vimalchand, Fluid Phase Equil., 40, 185 (1988). R. L. Cotterman, B. J. Schwarz, J. M. Prausnitz, AIChE-J., 32, 1787 (1986). D. D. Liu, J. M. Prausnitz, Ind. Eng. Chem. Process Des. Dev., 19, 205 (1980). M. Ohzono, Y. Iwai, Y. Arai, J. Chem. Eng. Japan, 17, 550 (1984). S. Itsuno, Y. Iwai, Y. Arai, Kagaku Kogaku Ronbunshu, 12, 349 (1986). Y. Iwai, Y. Arai, J. Japan Petrol. Inst., 34, 416 (1991). C. H. Chien, R. A. Greenkorn, K. C. Chao, AIChE-J., 29, 560 (1983). T. Boublik, I. Nezbeda, Chem. Phys. Lett., 46, 315 (1977). H. Masuoka, K. C. Chao, Ind. Eng. Chem. Fundam., 23, 24 (1984). M. T. Raetzsch, E. Regener, Ch. Wohlfarth, Acta Polymerica, 37, 441 (1986). E. Regener, Ch. Wohlfarth, M. T. Raetzsch, Acta Polymerica, 37, 499, 618 (1986). E. Regener, Ch. Wohlfarth, M. T. Raetzsch, S. Hoering, Acta Polymerica, 39, 618 (1988). Ch. Wohlfarth, E. Regener, Plaste & Kautschuk, 35, 252 (1988). Ch. Wohlfarth, Plaste & Kautschuk, 37, 186 (1990). Ch. Wohlfarth, B. Zech, Makromol. Chem., 193, 2433 (1992). U. Finck, T. Heuer, Ch. Wohlfarth, Ber. Bunsenges. Phys. Chem., 96, 179 (1992). Ch. Wohlfarth, U. Finck, R. Schultz, T. Heuer, Angew. Makromol. Chem., 198, 91 (1992). Ch. Wohlfarth, Acta Polymerica, 43, 295 (1992). Ch. Wohlfarth, Plaste & Kautschuk, 39, 367 (1992). Ch. Wohlfarth, J. Appl. Polym. Sci., 48, 1923 (1993). Ch. Wohlfarth, Plaste & Kautschuk, 40, 272 (1993). Ch. Wohlfarth, Plaste & Kautschuk, 41, 163 (1994). Ch. Wohlfarth, Macromol. Chem. Phys., 198, 2689 (1997). J. C. Pults, R. A. Greenkorn, K. C. Chao, Chem. Eng. Sci., 44, 2553 (1989). J. C. Pults, R. A. Greenkorn, K. C. Chao, Fluid Phase Equil., 51, 147 (1989). R. Sy-Siong-Kiao, J. M. Caruthers, K. C. Chao, Ind. Eng. Chem. Res., 35, 1446 (1996). C. R. Novenario, J. M. Caruthers, K. C. Chao, Fluid Phase Equil., 142, 83 (1998). C. R. Novenario, J. M. Caruthers, K. C. Chao, Ind. Eng. Chem. Res., 37, 3142 (1998). C. R. Novenario, J. M. Caruthers, K. C. Chao, J. Appl. Polym. Sci., 67, 841 (1998). S. H. Huang, M. Radosz, Ind. Eng. Chem., 29, 2284 (1990). S. S. Chen, A. Kreglewski, Ber. Bunsenges. Phys. Chem., 81, 1048 (1977). M. Banaszak, Y. C. Chiew, M. Radosz, Phys. Rev. E, 48, 3760 (1993). M. Banaszak, Y. C. Chiew, R. O’Lenick, M. Radosz, J. Chem. Phys., 100, 3803 (1994). M. Banaszak, C. K. Chen, M. Radosz, Macromolecules, 29, 6481 (1996). K. P. Shukla, W. G. Shapman, Mol. Phys., 91, 1075 (1997). H. Adidharma, M. Radosz, Ind. Eng. Chem. Res., 37, 4453 (1998). S.-J. Chen, M. Radosz, Macromolecules, 25, 3089 (1992). S.-J. Chen, I. G. Economou, M. Radosz, Macromolecules, 25, 4987 (1992). C. J. Gregg, S.-J. Chen, F. P. Stein, M. Radosz, Fluid Phase Equil., 83, 375 (1993). S.-J. Chen, M. Banaszak, M. Radosz, Macromolecules, 28, 1812 (1995). P. Condo, M. Radosz, Fluid Phase Equil., 117, 1 (1996). K. L. Albrecht, F. P. Stein, S. J. Han, C. J. Gregg, M. Radosz, Fluid Phase Equil., 117, 117 (1996). C. Pan, M. Radosz, Ind. Eng. Chem. Res., 37, 3169 (1998) and 38, 2842 (1999). G. Sadowski, L. V. Mokrushina, W. Arlt, Fluid Phase Equil., 139, 391 (1997). J. Groß, G. Sadowski, submitted to Fluid Phase Equil. C. F. Kirby, M. A. McHugh, Chem. Rev., 99, 565 (1999). R. B. Gupta, J. M. Prausnitz, Ind. Eng. Chem. Res., 35, 1225 (1996). T. Hino, Y. Song, J. M. Prausnitz, Macromolecules, 28, 5709, 5717, 5725 (1995). S. M. Lambert, Song, J. M. Prausnitz, Macromolecules, 28, 4866 (1995). G. M. Kontogeorgis, V. I. Harismiadis, A. Fredenslund, D. P. Tassios, Fluid Phase Equil., 96, 65 (1994). V. I. Harismiadis, G. M. Kontogeorgis, A. Fredenslund, D. P. Tassios, Fluid Phase Equil., 96, 93 (1994). T. Sako, A. W. Hu, J. M. Prausnitz, J. Appl. Polym. Sci., 38, 1839 (1989).


223

378 G. Soave, Chem. Eng. Sci., 27, 1197 (1972). 379 V. I. Harismiadis, G. M. Kontogeorgis, A. Saraiva, A. Fredenslund, D. P. Tassios, Fluid Phase Equil., 100, 63 (1994). 380 N. Orbey, S. I. Sandler, AIChE-J., 40, 1203 (1994). 381 R. Stryjek, J. H. Vera, Can. J. Chem. Eng., 64, 323 (1986). 382 D. S. H. Wong, S. I. Sandler, AIChE-J., 38, 671 (1992). 383 C. Zhong, H. Masuoka, Fluid Phase Equil., 126, 1 (1996). 384 C. Zhong, H. Masuoka, Fluid Phase Equil., 126, 59 (1996). 385 C. Zhong, H. Masuoka, Fluid Phase Equil., 144, 49 (1998). 386 H. Orbey, C.-C. Chen, C. P. Bokis, Fluid Phase Equil., 145, 169 (1998). 387 H. Orbey, C. P. Bokis, C.-C. Chen, Ind. Eng. Chem. Res., 37, 1567 (1998).

APPENDIX 4.4A Table of polymer-solvent systems for which experimental VLE-data have been reported in the literature (the references are given at the end of this table) Solvent

T, K

Ref.

Solvent

T, K

Ref.

acrylonitrile/butadiene copolymer acetonitrile

333.15 121

n-hexane

333.15 121

chloroform

333.15 121

n-octane

333.15 121

cyclohexane

333.15 121

n-pentane

333.15 121

acrylonitrile/styrene copolymer benzene

343.15 130, 134

o-xylene

398.15 130, 134

1,2-dichloroethane

343.15 121

m-xylene

398.15 130, 134

1,2-dichloroethane

353.15 121

p-xylene

373.15 130, 134

propylbenzene

398.15 130, 134

p-xylene

398.15 130, 134

toluene

343.15 130, 134

p-xylene

423.15 130, 134

toluene

373.15 130, 134 p-bromostyrene/p-methylstyrene copolymer

toluene

293.20 72 cellulose acetate

acetone

303.15 81

1,4-dioxane

308.15 81

acetone

308.15 81

methyl acetate

303.15 81

N,N-dimethylformamide

322.85 75

methyl acetate

308.15 81


342.55 75

pyridine

303.15 81

1,4-dioxane

303.15 81

pyridine

308.15 81

cellulose triacetate chloroform

303.15 77

dichloromethane

293.15 77

chloroform

308.15 77

dichloromethane

298.15 77

224

Christian Wohlfarth

Solvent

T, K

Ref.

Solvent

T, K

Ref.

dextran 111, 112, 119, 154

water

293.15

water

298.15 76

water

313.15 111

water

333.15 111, 119

di(trimethylsilyl)-poly(propylene oxide) toluene

323.15 75

toluene

342.65 75

n-decane

342.45 75

ethylene/vinyl acetate copolymer benzene

303.15 36

n-propyl acetate

343.15 37

benzene

323.15 36

n-propyl acetate

363.15 37

benzene

328.15 36

toluene

303.15 36

benzene

333.15 35

toluene

323.15 36

benzene

343.15 36

toluene

333.15 35

benzene

353.15 35

toluene

343.15 36

benzene

373.15 34, 35

toluene

353.15 35

butyl acetate

323.15 37

toluene

363.15 36

butyl acetate

343.15 37

toluene

373.15 34, 35

butyl acetate

363.15 37

o-xylene

323.15 36

chloroform

333.15 121

o-xylene

343.15 36

cyclohexane

353.15 121

o-xylene

363.15 36

ethyl acetate

303.25 37

p-xylene

323.15 36

ethyl acetate

323.15 37

p-xylene

333.15 35

ethyl acetate

343.15 37

p-xylene

343.15 36

methyl acetate

303.15 37

p-xylene

353.15 35

methyl acetate

323.15 37

p-xylene

363.15 36

n-propyl acetate

303.15 37

p-xylene

373.15 34, 35

n-propyl acetate

323.15 37 hydroxypropyl cellulose

acetone

298.15 54

tetrahydrofuran

298.15 54

ethanol

298.15 54

water

298.15 141

hydroxypropyl starch water

293.15 117

water

298.15 160


Solvent

225

T, K

Ref.

Solvent

T, K

Ref.

natural rubber acetone

273.15 25

2-butanone

318.15 25

acetone

298.15 25

ethyl acetate

298.15 25

benzene

298.15 6

ethyl acetate

323.15 25

2-butanone

298.15 25

toluene

303.00 82

nitrocellulose acetone

293.00 79

ethyl propyl ether

293.00 79

acetone

303.15 81

methyl acetate

303.15 81

acetone

308.15 81

methyl acetate

308.15 81

acetonitrile

293.00 79

3-methyl-2-butanone

293.00 79

ethyl formate

293.15 78

3-methylbutyl acetate

293.15 78

cyclopentanone

293.00 79

nitromethane

293.00 79

3,3-dimethyl-2-butanone

293.00 79

2-pentanone

293.00 79

2,4-dimethyl-3-pentanone 293.00 79

ethyl propionate

293.15 78

1,4-dioxane

293.00 79

propyl acetate

293.15 78

ethyl acetate

293.15 78 nylon 6,6

water

296.15 103 nylon 6,10

water

296.15 103 poly(acrylic acid)

ethanol

303.15 145

water

303.15 145

poly(acrylonitrile) 1,2-dichloroethane

353.15 121


323.25 75


343.55 75

polyamidoamine dendrimers acetone

308.15 144

methanol

308.15 144

acetonitrile

313.15 144

1-propylamine

308.15 144

chloroform

308.15 144 poly(benzyl ether) dendrimers

acetone

323.15 144

tetrahydrofuran

343.15 144

chloroform

323.15 144

toluene

343.15 144

chloroform

343.15 144

n-pentane

313.15 144

226

Christian Wohlfarth

Solvent cyclohexane

T, K

Ref.

Solvent

T, K

Ref.

333.15 144 poly(γ-benzyl-L-glutamate)

chloroform

303.15 51 poly(p-bromostyrene)

toluene

293.20 72 polybutadiene

benzene

296.65 7

ethylbenzene

403.15 106

chloroform

296.65 7

n-hexane

296.65 7

chloroform

298.15 69

n-hexane

333.15 121

chloroform

333.15 121

n-nonane

353.15 106

cyclohexane

296.65 7

n-nonane

373.15 106

cyclohexane

333.15 121

n-nonane

403.15 106

dichloromethane

296.65 7

n-pentane

333.15 121

ethylbenzene

353.15 106

tetrachloromethane

296.65 7

ethylbenzene

373.15 106

toluene

296.65 7

poly(n-butyl acrylate) benzene

296.65 61

tetrachloromethane

296.65 61

chloroform

296.65 61

toluene

296.65 61

dichloromethane

296.65 61 poly(n-butyl methacrylate)

benzene

323.65 75

mesitylene

373.15 75

benzene

343.45 75

mesitylene

403.15 75

2-butanone

323.65 75

3-pentanone

323.55 75

2-butanone

344.45 75

3-pentanone

343.95 75

chloroform

323.75 75

propylbenzene

344.35 75

chloroform

343.75 121

tetrachloromethane

323.65 75

cumene

373.15 75

tetrachloromethane

344.45 75

cumene

403.15 75

toluene

323.35 75

cyclohexane

308.15 125

toluene

343.75 75

cyclohexane

318.15 125

toluene

373.15 75

cyclohexane

328.15 125

o-xylene

344.45 75

cyclohexane

338.15 125

o-xylene

373.15 75

diethyl ether

298.15 159

o-xylene

403.15 75


Solvent

227

T, K

Ref.

Solvent

T, K

1,2-dichloroethane

323.95 75

m-xylene

343.95 75

1,2-dichloroethane

343.15 75

m-xylene

373.15 75


323.65 75

m-xylene

403.15 75


344.45 75

p-xylene

344.45 75

ethylbenzene

343.75 75

p-xylene

373.15 75

ethylbenzene

373.15 75

p-xylene

403.15 75

ethylbenzene

403.15 75

Ref.

poly(tert-butyl methacrylate) benzene

323.15 75

3-pentanone

342.65 75

benzene

342.65 75

propylbenzene

342.65 75

2-butanone

323.15 75

tetrachloromethane

323.15 75

2-butanone

342.65 75

tetrachloromethane

342.65 75

chloroform

323.15 75

toluene

323.15 75

cumene

373.15 75

toluene

342.75 75

1,2-dichloroethane

323.15 75

toluene

373.15 75

1,2-dichloroethane

342.65 75

o-xylene

342.65 75


323.15 75

o-xylene

373.15 75


342.65 75

m-xylene

342.65 75

ethylbenzene

342.65 75

m-xylene

373.15 75

ethylbenzene

373.35 75

p-xylene

342.65 75

mesitylene

373.15 75

p-xylene

373.15 75

3-pentanone

323.15 75 poly(ε-caprolacton)

tetrachloromethane

338.15 65 polycarbonate-bisphenol-A

chlorobenzene

413.15 123, 139

mesitylene

453.15 139

chlorobenzene

433.15 139

n-pentane

303.15 145

chlorobenzene

453.15 139

toluene

413.15 139

ethanol

303.15 145

water

303.15 145

ethylbenzene

413.15 139

m-xylene

413.15 139

ethylbenzene

433.15 139

m-xylene

453.15 139

mesitylene

413.15 139

p-xylene

413.15 139

mesitylene

433.15 139

228

Christian Wohlfarth

Solvent

T, K

Ref.

Solvent

T, K

Ref.

poly(o-chlorostyrene) benzene

298.15 115

benzene

313.15 115

2-butanone

313.15 115

poly(p-chlorostyrene) toluene

293.20 72 polydecene

toluene

303.15 3 poly(dimethyl siloxane)

benzene

298.15 52, 62

benzene

303.00 63, 66, 101 n-nonane

benzene

313.15 62

octamethyl cyclotetrasiloxane

413.15 85

2-butanone

303.15 53

n-octane

298.15 62

chloroform

303.00 99

n-octane

303.15 157

cyclohexane

293.15 161

n-octane

313.15 62

cyclohexane

303.15 63, 66, 161 n-pentane

303.15 63, 147

dichloromethane

303.00 99

n-pentane

313.15 145

n-heptane

298.15 62

toluene

298.15 62

n-heptane

303.15 63, 157

toluene

308.15 124

n-heptane

313.15 62

toluene

313.15 62

n-hexane

298.09 66

toluene

318.15 124

n-hexane

303.15

toluene

328.15 124

n-hexane

308.08 66

2,2,4-trimethylpentane

298.15 62

n-hexane

313.15 145


313.15 62

benzene

303.15 105

63, 66, 100, 147

hexamethyl disiloxane

298.15 62 313.15 157

poly(1,3-dioxolane) benzene

313.15 105

polydodecene toluene

303.15 3 poly(ethyl acrylate)

benzene

296.65 61

tetrachloromethane

296.65 61

chloroform

296.65 61

toluene

296.65 61

dichloromethane

296.65 61


Solvent

229

T, K

Ref.

Solvent

T, K

Ref.

polyethylene chlorobenzene

393.15 109

propyl acetate

426.15 137

chlorobenzene

403.15 109

propyl acetate

474.15 137

chlorobenzene

413.15 109

2-propylamine

427.15 137

cyclopentane

425.65 137

2-propylamine

475.15 137

cyclopentane

474.15 137

toluene

393.15 109

ethylbenzene

413.15 116

o-xylene

413.15 116

n-heptane

382.05 2

m-xylene

413.15 116

n-pentane

423.65 137

p-xylene

353.15 1

n-pentane

474.15 137

p-xylene

363.15 1, 35

3-pentanol

423.15 137

p-xylene

373.15 35

3-pentanol

473.15 137

p-xylene

383.15 35

3-pentanone

425.15 137

p-xylene

403.15 116

3-pentanone

477.15 137

p-xylene

413.15 116

1-pentene

423.65 137

p-xylene

423.15 116

1-pentene

474.15 137 poly(ethylene glycol)

benzene

297.75 43

1-propanol

323.15 45

benzene

307.75 43

1-propanol

333.15 94

benzene

313.15 42

1-propanol

343.15 45

benzene

323.15 42

1-propanol

353.15 94

benzene

343.15 42

1-propanol

373.15 45

1-butanol

323.15 42

tetrachloromethane

303.15 95, 96

1-butanol

343.15 42

toluene

323.15 42, 75

1-butanol

373.15 42

toluene

343.15 42, 75

1-butanol

403.15 42

toluene

373.15 42

chloroform

323.15 146

water

293.15

chloroform

333.15 146

water

298.15

104, 111, 112, 117, 119 39, 40, 41, 76, 128, 129, 158, 160

ethanol

303.15 45, 157

water

303.15 38

ethanol

313.15 45

water

308.15 40, 151

ethanol

323.15 45

water

313.15 104, 111

230

Christian Wohlfarth

Solvent

T, K

Ref.

Solvent

T, K

Ref.

ethylbenzene

323.15 42, 44

water

318.15 151

ethylbenzene

343.15 42, 44

water

323.15 38, 108

ethylbenzene

343.75 75

water

328.15 38, 151

ethylbenzene

373.15 42, 44

water

333.15

ethylbenzene

403.15 44

water

338.15 38, 151

1-hexanol

323.15 46

water

343.15 108

1-hexanol

373.15 46

p-xylene

323.15 45

1-hexanol

403.15 46

p-xylene

343.15 45

methanol

303.15 157

p-xylene

373.15 45

1-propanol

303.15 45

p-xylene

403.15 45

38, 104, 108, 111, 119, 146

poly(ethylene glycol) dimethyl ether chloroform

278.68 80

tetrachloromethane

278.68 80

tetrachloromethane

303.15 96

poly(ethylene glycol) monomethyl ether tetrachloromethane

303.15 96 poly(ethylene oxide)

acetone

323.15 152

benzene

423.55 47

acetone

353.15 136

2-butanone

353.15 136

benzene

318.85 74

chloroform

298.15 48

benzene

323.45 74

chloroform

323.15 152

benzene

328.15 65

chloroform

343.15 152

benzene

343.15 65, 74

chloroform

333.15 121

benzene

348.25 47

cyclohexane

353.15 136

benzene

353.15 136

toluene

353.15 136

benzene

361.25 47

toluene

372.98 68

benzene

375.15 47

p-xylene

353.15 136

benzene

398.85 47 poly(ethylene oxide)-b-poly(tert-butyl methacrylate) diblock copolymer

toluene

323.15 75

toluene

343.75 75

poly(ethylene oxide)-b-poly(methyl methacrylate) diblock copolymer toluene

323.41 68

toluene

373.27 68


Solvent toluene

231

T, K

Ref.

Solvent

T, K

Ref.

343.27 68 poly(ethylene oxide)-b-poly(methyl methacrylate)-b-poly(ethylene oxide) triblock copolymer

toluene

323.08 68

toluene

373.26 68

toluene

343.17 68

poly(ethylene oxide)-b-poly(propylene oxide) diblock copolymer tetrachloromethane

303.15 96

poly(ethylene oxide)-b-poly(propylene oxide)-b-poly(ethylene oxide) triblock copolymer tetrachloromethane

303.15 96

toluene

323.35 75

toluene

343.75 75

poly(ethylene oxide)-b-polystyrene-b-poly(ethylene oxide) triblock copolymer toluene

323.35 75

toluene

343.75 75

poly(ethyl methacrylate) benzene

296.65 61

tetrachloromethane

296.65 61

chloroform

296.65 61

toluene

296.65 61

dichloromethane

296.65 61 polyheptene

toluene

303.15 3 poly(4-hydroxystyrene)

acetone

293.15 97

acetone

308.15 97

acetone

298.15 97

acetone

313.15 97

acetone

303.15 97

acetone

318.15 97

polyisobutylene benzene

298.15 57, 8, 107

ethylbenzene

338.15 107

benzene

300.05 9

n-heptane

296.65 60

benzene

313.15

n-heptane

338.15 156

benzene

333.20 70

n-hexane

298.15 107

benzene

338.15 8, 107

n-hexane

313.15 107

benzene

353.20 70

n-hexane

338.15 107, 156

n-butane

298.15 4

2-methylbutane

298.15 4

n-butane

308.15 4

2-methylbutane

308.15 4

n-butane

319.65 4

2-methylbutane

319.65 4

chloroform

296.65 60

2-methylpropane

308.15 4

8, 57, 70, 107

232

Christian Wohlfarth

Solvent

T, K

Ref.

cyclohexane

298.15

8, 58, 60, 107, 147

cyclohexane

Solvent

T, K

Ref.

2-methylpropane

319.65 4

308.15 125

n-nonane

338.15 156

cyclohexane

313.15 70, 107

n-octane

338.15 156

cyclohexane

315.15 8

n-pentane

298.15 4, 5, 59, 65

cyclohexane

318.15 125

n-pentane

308.15 4, 5

cyclohexane

328.15 125

n-pentane

318.15 5

cyclohexane

333.20 70

n-pentane

319.65 4

cyclohexane

338.15 8, 107, 125 n-pentane

328.15 5

cyclopentane

296.65 60

propane

308.15 4

2,2-dimethylbutane

296.65 60

tetrachloromethane

296.65 60

2,2-dimethylpropane

298.15 4

toluene

298.15 107

2,2-dimethylpropane

308.15 4

toluene

313.15 107

2,2-dimethylpropane

319.65 4

toluene

338.15 107

ethylbenzene

298.15 107


296.65 60

ethylbenzene

313.15 107 1,4-cis-polyisoprene

benzene

296.65 7

dichloromethane

296.65 7

benzene

353.15 10

tetrachloromethane

296.65 7

chloroform

296.65 7

toluene

296.65 7

cyclohexane

296.65 7 polyisoprene, hydrogenated

cyclohexane

323.15 161 poly(maleic anhydride)

acetone

323.15 140

methanol

333.15 140

poly(methyl acrylate) benzene

296.65 61

tetrachloromethane

296.65 61

chloroform

296.65 61

toluene

296.65 61

dichloromethane

296.65 61 poly(methyl methacrylate)

acetone

308.15 131, 150


343.45 56

acetone

323.15 133

ethyl acetate

308.15 131

benzene

296.65 61

ethylbenzene

398.15 153


Solvent

233

T, K

Ref.

Solvent

T, K

Ref.

benzene

323.15 56

mesitylene

403.15 67

benzene

343.15 56

methyl acetate

323.15 133

2-butanone

308.15 132, 150

tetrachloromethane

323.15 56

2-butanone

323.15 23, 56

tetrachloromethane

343.75 56

2-butanone

343.55 56

toluene

296.65 61

chloroform

296.65 61

toluene

323.15 23, 56, 75

chloroform

303.15 146

toluene

343.15 56, 75

chloroform

308.15 132

toluene

373.97 68

chloroform

323.15

toluene

433.15 19

cyclohexanone

323.15 146

p-xylene

323.15 56

dichloromethane

296.65 61

p-xylene

343.15 56

1,2-dichloroethane

323.15 67

p-xylene

373.15 56

1,2-dichloroethane

343.15 67

p-xylene

403.15 56


323.35 56

p-xylene

409.35 56

56, 133, 142, 146

poly(α-methylstyrene) cumene

338.15 83

tetrahydrofuran

298.15 92

1,4-dioxane

313.15 89

toluene

298.15 28, 90

α-methylstyrene

303.15 90

toluene

303.15 90

α-methylstyrene

308.15 90

toluene

308.15 90

α-methylstyrene

313.15 90

toluene

313.15 90

α-methylstyrene

338.15 83 poly(p-methylstyrene)

toluene

293.20 72 polyoctadecene

toluene

303.15 3 polypropylene


3-pentanone

318.15 11


tetrachloromethane

298.15 84

3-pentanone

298.15 11 poly(propylene glycol)

n-decane

343.45 75

methanol

298.15 49

234

Christian Wohlfarth

Solvent

T, K

Ref.

Solvent

T, K

Ref.

ethylbenzene

342.65 75

tetrachloromethane

303.15 96

n-hexane

298.10 87

toluene

323.15 75

n-hexane

312.65 87

toluene

343.75 75

n-hexane

323.15 87

water

298.15 41, 87

methanol

263.15 49

water

303.15 38

methanol

273.15 49

water

312.65 87

methanol

288.15 49

water

323.15 38

poly(propylene glycol) dimethyl ether chloroform

278.68 50

tetrachloromethane

278.68 50

poly(propylene imine) dendrimers acetone

323.15 155

n-heptane

348.15 155

acetonitrile

343.15 155

n-hexane

338.15 155

acetonitrile

348.15 155

n-nonane

338.15 155

chloroform

323.15 155

n-octane

338.15 155

chloroform

343.15 155

tetrahydrofuran

323.15 155

n-heptane

338.15 155

toluene

343.15 155

n-heptane

343.15 155

triethylamine

338.15 155

poly(propylene oxide) benzene

298.15 147

methanol

298.15 147

benzene

320.35 73

methanol

303.15 157

benzene

333.35 73

methanol

313.15 145

benzene

343.05 73

propanol

303.15 157

benzene

347.85 73

water

303.15 157

ethanol

303.15 157 poly(propylene oxide)-b-poly(ethylene oxide) diblock copolymer

ethylbenzene

343.75 75

poly(propylene oxide)-b-poly(ethylene oxide)-b-poly(propylene oxide) triblock copolymer toluene

323.05 75

toluene

342.65 75

polystyrene acetone

298.15 27

dichloromethane

296.65 17

acetone

323.15 27, 152

1,4-dioxane

293.15 16

acetone

393.15 122

1,4-dioxane

323.15 23, 26

acetone

423.15 122

dipropyl ether

293.15 16


Solvent

235

T, K

Ref.

Solvent

T, K

Ref.

acetonitrile

393.15 122

ethyl acetate

313.15 146

acetonitrile

423.15 122

ethyl acetate

333.15 146

anisole

323.15 26

ethylbenzene

303.15 15

benzene

288.15 102

ethylbenzene

323.15 22

benzene

293.15 16

ethylbenzene

343.15 22

benzene

296.65 17

ethylbenzene

398.15 153

benzene

298.15 65

ethylbenzene

403.15 19

benzene

303.15

ethylbenzene

413.15 19

benzene

313.15 149

ethylbenzene

433.15 19

benzene

318.15 102

ethylbenzene

443.15 19

benzene

323.15 12, 138

ethylbenzene

451.15 19

benzene

333.15 102, 71

n-hexane

393.15 122

benzene

343.15 12

n-hexane

423.15 122

benzene

353.20 71

methyl acetate

323.15 133

benzene

393.15 122

n-nonane

403.15 106

benzene

403.15 19

n-nonane

423.15 106

benzene

423.15 122

n-nonane

448.15 106

benzene

428.15 19

3-pentanone

293.15 16

2-butanone

298.15 24, 26

propyl acetate

298.15 27

2-butanone

321.65 23

propyl acetate

343.14 27

2-butanone

343.15 24

tetrachloromethane

293.15 16

2-butanone

393.15 122

tetrachloromethane

296.65 17

2-butanone

423.15 122

toluene

293.15 16

butyl acetate

308.15 159

toluene

296.65 17

butyl acetate

323.15 26

toluene

298.15 24, 102, 138

tert-butyl acetate

283.15 64

toluene

303.15 14, 15, 20

tert-butyl acetate

303.15 64

toluene

308.15 127

tert-butyl acetate

323.15 64

toluene

313.15 149

tert-butyl acetate

343.15 64

toluene

321.65 23

tert-butyl acetate

363.15 64

toluene

323.15

chloroform

296.65 17

toluene

333.15 24, 71

chloroform

298.15 27, 65, 138 toluene

12, 71, 102, 149

14, 20, 71, 138, 149

343.15 20

236

Christian Wohlfarth

Solvent

T, K

Ref.

Solvent

T, K

Ref.

chloroform

323.15 27, 138, 152 toluene

353.20 71

cyclohexane

293.15 16

toluene

373.15 20, 21, 123

cyclohexane

296.65 17

toluene

383.15 19

cyclohexane

297.15 13

toluene

393.15 21, 122, 123

cyclohexane

303.15

toluene

403.15 19

cyclohexane

308.15 13, 125

toluene

413.15 19

cyclohexane

313.15 71, 108, 149 toluene

423.15 122

cyclohexane

318.15 13, 125

1,2,4-trimethylbenzene

443.15 19

cyclohexane

323.15

o-xylene

323.15 26

cyclohexane

328.15 125

o-xylene

373.15 21

cyclohexane

333.15 71, 108

o-xylene

403.15 21

cyclohexane

343.15 12

m-xylene

323.15 146

cyclohexane

353.20 71

m-xylene

373.15 21

cyclohexane

338.15 125

m-xylene

403.15 106

cyclohexane

393.15 122

m-xylene

423.15 106

cyclohexane

423.15 122

m-xylene

448.15 106

cyclohexanone

313.15 146

p-xylene

373.15 21

cyclohexanone

333.15 146

p-xylene

393.15 122

1,2-dichloroethane

343.15 121

p-xylene

403.15 21

1,2-dichloroethane

353.15 121

p-xylene

423.15 122

12, 14, 15, 108, 149

12, 14, 108, 152

polystyrene-b-polybutadiene-b-polystyrene triblock copolymer cyclohexane

323.15 161

cyclohexane

373.15 161

cyclohexane

348.15 161

cyclohexane

393.15 161

polystyrene-b-poly(ethylene oxide) diblock copolymer toluene

322.95 75

toluene

342.65 75

polystyrene-b-polyisoprene-b-polystyrene triblock copolymer cyclohexane

323.15 161 polystyrene-b-poly(methyl methacrylate) diblock copolymer

benzene

343.15 153

ethylbenzene

398.15 153

1,4-dimethylbenzene

398.15 153

toluene

343.15 153

ethylbenzene

373.15 153


398.15 153


Solvent

237

T, K

Ref.

Solvent

T, K

Ref.

poly(tetramethylene oxide) benzene

318.15 91

1,4-dioxane

303.15 55

tetrahydrofuran

318.15 91

1,4-dioxane

313.15 55

poly(vinyl acetate) acetone

298.15 110

1-chloropropane

313.15 30

acetone

303.15 29, 30

1,2-dichloroethane

299.55 86

acetone

308.15 110

ethyl acetate

303.15 29

acetone

313.15 30

methanol

303.15 29

acetone

318.15 110

methanol

353.15 121

acetone

323.15 30

1-propanol

323.15 30

allyl chloride

313.15 30

1-propylamine

313.15 30

benzene

303.15 29, 30, 31

2-propylamine

313.15 30

benzene

313.15 98

toluene

299.55 86

benzene

323.15 30

toluene

308.15 19

benzene

333.15 98

toluene

313.15 19, 98, 120

1-butanol

353.15 121

toluene

333.15 98, 120

chloroform

308.15 19

toluene

353.15 120

chloroform

313.15 19

vinyl acetate

303.15 31

chloroform

333.15 121 poly(vinyl alcohol)

water

303.15 32, 145, 147 poly(vinylcarbazol)

benzene

279.15 33

benzene

308.15 33

benzene

288.15 33

benzene

318.15 33

benzene

298.15 33

benzene

328.15 33

poly(vinyl chloride) 2-butanone

333.15 146

tetrachloromethane

338.15 65

cyclohexanone

313.15 146

tetrahydrofuran

315.65 23

cyclohexanone

333.15 146

toluene

316.35 23

dibutyl ether

315.35 23

vinyl chloride

340.15 93

1,4-dioxane

315.65 23 poly(vinyl methyl ether)

benzene

298.15 65

ethylbenzene

398.15 118

238

Christian Wohlfarth

Solvent

T, K

Ref.

Solvent

T, K

Ref.

benzene

323.15 75

propylbenzene

373.15 75

benzene

343.15 75

toluene

323.15 75

chlorobenzene

343.15 75

toluene

343.15 75

chlorobenzene

373.15 75

o-xylene

363.15 75

chloroform

298.15 65

o-xylene

373.15 118

cyclohexane

308.15 125

o-xylene

398.15 118

cyclohexane

318.15 125

m-xylene

373.15 118

cyclohexane

328.15 125

m-xylene

398.15 118

cyclohexane

338.15 125

p-xylene

373.15 118

ethylbenzene

343.15 75

p-xylene

398.15 118

ethylbenzene

373.15 118 starch

water

353.15 143

water

383.15 143

water

363.15 143

water

393.15 143

water

373.15 143

water

403.15 143

styrene/butadiene copolymer acetone

323.15 121

ethylbenzene

403.15 113

acetone

333.15 121

n-hexane

343.15 121

benzene

343.15 126, 134

mesitylene

398.15 126, 134

chloroform

323.15 121

n-nonane

373.15 114

cyclohexane

296.65 121

n-nonane

403.15 114

cyclohexane

333.15 121

n-pentane

333.15 121

cyclohexane

343.15 126, 134

toluene

343.15 126, 134

ethylbenzene

373.15

toluene

373.15 126, 134

ethylbenzene

398.15 126, 134

p-xylene

398.15 126, 134

113, 126, 134

styrene/butyl methacrylate copolymer acetone

333.15 121

chloroform

343.15 121

styrene/docosyl maleate copolymer acetone

323.15 140

cyclohexane

333.15 140

acetone

343.15 140

methanol

333.15 140

styrene/dodecyl maleate copolymer acetone

323.15 140

cyclohexane

333.15 140

acetone

343.15 140

methanol

333.15 140


Solvent

T, K

239

Ref.

Solvent

T, K

Ref.

styrene/maleic anhydride copolymer acetone

323.15 140

methanol

333.15 140

styrene/methyl methacrylate copolymer acetone

323.15 133

mesitylene

398.15 135

benzene

343.15 135

methyl acetate

323.15 133

chloroform

323.15 133

toluene

343.15 135

ethylbenzene

373.15 135

toluene

373.15 135

ethylbenzene

398.15 135

p-xylene

398.15 135

styrene/pentyl maleate copolymer acetone

323.15 140

cyclohexane

333.15 140

acetone

343.15 140

methanol

333.15 140

vinyl acetate/vinyl chloride copolymer benzene

398.15 148

ethylbenzene

428.15 148

benzene

418.15 148

n-octane

398.15 148

1-butanol

353.15 121

n-octane

418.15 148

chlorobenzene

398.15 148

p-xylene

398.15 148

ethylbenzene

398.15 148

p-xylene

418.15 148

ethylbenzene

418.15 148

REFERENCE LIST OF APPENDIX 4.4A 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

G. Krahn, Dissertation, TH Leuna-Merseburg, 1973*). J. H. Van der Waals, J. J. Hermans, Rec. Trav. Chim. Pays-Bas, 69, 971 (1950). P. J. T. Tait, P. J. Livesey, Polymer, 11, 359 (1970). S. Prager, E. Bagley, F. A. Long, J. Amer. Chem. Soc., 75, 2742 (1953). C. H. Baker, W. B. Brown, G. Gee, J. S. Rowlinson, D. Stubley, R. E. Yeadon, Polymer, 3, 215 (1962). P. E. Eichinger, P. J. Flory, Trans. Faraday Soc., 64, 2035 (1968). S. Saeki, J. C. Holste, D. C. Bonner, J. Polym. Sci., Polym. Phys. Ed., 20, 793 (1982). C. E. H. Bawn, R. D. Patel, Trans. Faraday Soc., 52, 1664 (1956). R. S. Jessup, J. Res. Natl. Bur. Stand., 60, 47 (1958). D. C. Bonner, J. M. Prausnitz, J. Polym. Sci., Polym. Phys. Ed., 12, 51 (1974). W. B. Brown, G. Gee, W. D. Taylor, Polymer, 5, 362 (1964). M. T. Raetzsch, M. Opel, Ch. Wohlfarth, Acta Polymerica, 31, 217 (1980). W. R. Krigbaum, D. O. Geymer, J. Amer. Chem. Soc., 81, 1859 (1959). K. Schmoll, E. Jenckel, Ber. Bunsenges. Phys. Chem., 60, 756 (1956). T. Katayama, K. Matsumara, Y. Urahama, Kagaku Kogaku, 35, 1012 (1971). E. C. Baughan, Trans. Faraday Soc., 44, 495 (1948). S. Saeki, J. C. Holste, D. C. Bonner, J. Polym. Sci., Polym. Phys. Ed., 19, 307 (1981). M. Opel, Dissertation, TH Leuna-Merseburg, 1978*). J. S. Vrentas, J. L. Duda, S. T. Hsieh, Ind. Eng. Chem., Process Des. Dev., 22, 326 (1983). M. Braeuer, Dissertation, TH Leuna-Merseburg, 1983*). G. Illig, Dissertation, TH Leuna-Merseburg, 1981*). M. T. Raetzsch, G. Illig, Ch. Wohlfarth, Acta Polymerica, 33, 89 (1982). P. J. T. Tait, A. M. Abushihada, Polymer, 18, 810 (1977).

240

24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77

Christian Wohlfarth C. E. H. Bawn, R. F. J. Freeman, A. R. Kamaliddin, Trans. Faraday Soc., 46, 677 (1950). C. Booth, G. Gee, G. Holden, G. R. Williamson, Polymer, 5, 343 (1964). G. Illig, Diploma Paper, TH Leuna-Merseburg, 1973*). C. E. H. Bawn, M. A. Wajid, Trans. Faraday Soc., 52, 1658 (1956). I. Noda, N. Kato, T. Kitano, M. Nagasawa, Macromolecules, 14, 668 (1981). K. Matsumara, T. Katayama, Kagaku Kogaku, 38, 388 (1974). R. J. Kokes, A. R. Dipietro, F. A. Long, J. Amer. Chem. Soc., 75, 6319 (1953). A. Nakajima, H. Yamakawa, I. Sakurada, J. Polym. Sci., 35, 489 (1959). I. Sakurada, A. Nakajima, H. Fujiwara, J. Polym. Sci., 35, 497 (1959). K. Ueberreiter, W. Bruns, Ber. Bunsenges. Phys. Chemie, 68, 541 (1964). J. Belorussow, Diploma Paper, TH Leuna-Merseburg, 1973*). K. Peinze, Diploma Paper, TH Leuna-Merseburg, 1972*). M. Boblenz, D. Glindemann, Diplom Paper, TH Leuna-Merseburg, 1975*). D. Kiessling, Diploma Paper, TH Leuna-Merseburg, 1976*). G. N. Malcolm, J. S. Rowlinson, Trans. Faraday Soc., 53, 921 (1957). Z. Adamcova, Sci. Pap. Prag. Inst. Chem. Technol., N2, 63 (1976). M. L. Lakhanpal, K. S. Chhina, S. C. Sharma, Indian J. Chem., 6, 505 (1968). Z. N. Medved, P. P. Petrova, O. G. Tarakanov, Vysokomol. Soedin., Ser. B, 24, 674 (1982). Ch. Wohlfarth, M. T. Raetzsch, Acta Polymerica, 37, 86 (1986). M. L. Lakhanpal, H. G. Singh, S. C. Sharma, Indian J. Chem., 6, 436 (1968). Ch. Wohlfarth, W. Zschoch, M. T. Raetzsch, Acta Polymerica, 32, 616 (1981). Ch. Wohlfarth, H. Hahmann, M. T. Raetzsch, Acta Polymerica, 32, 674 (1982). E. Regener, Diploma Paper, TH Leuna-Merseburg, 1983*). S. H. Chang, D. C. Bonner, J. Appl. Polym. Sci., 19, 2457 (1975). G. Allen, C. Booth, G. Gee, M. N. Jones, Polymer, 5, 367 (1964). M. L. Lakhanpal, B. E. Conway, J. Polym. Sci., 46, 75 (1960). R. W. Kershaw, G. N. Malcolm, Trans. Faraday Soc., 64, 323 (1968). K. Kubo, K. Ogino, Polymer, 16, 629 (1975). M. J. Newing, Trans. Faraday Soc., 46, 613 (1950). A. Muramoto, Polymer, 23, 1311 (1982). J. S. Aspler, D. G. Gray, Polymer, 23, 43 (1982). S. C. Sharma, M. L. Lakhanpal, J. Polym. Sci., Polym. Phys. Ed., 21, 353 (1983). E. Regener, Ch. Wohlfarth, M. T. Raetzsch, S. Hoering, Acta Polymerica, 39, 618 (1988). P. E. Eichinger, P. J. Flory, Trans. Faraday Soc., 64, 2053 (1968). P. E. Eichinger, P. J. Flory, Trans. Faraday Soc., 64, 2061 (1968). P. E. Eichinger, P. J. Flory, Trans. Faraday Soc., 64, 2066 (1968). S. Saeki, J. C. Holste, D. C. Bonner, J. Polym. Sci., Polym. Phys. Ed., 20, 805 (1982). S. Saeki, J. Holste, D. C. Bonner, J. Polym. Sci., Polym. Phys. Ed., 21, 2049 (1983). E. Dolch, M. Glaser, A. Heintz, H. Wagner, R. N. Lichtenthaler, Ber. Bunsenges. Phys. Chem., 88, 479 (1984). A. J. Ashworth, C.-F. Chien, D. L. Furio, D. M. Hooker, M. M. Kopecni, R. J. Laub, G. J. Price, Macromolecules, 17, 1090 (1984). K. Schotsch, B. A. Wolf, Makromol. Chem., 185, 2161 (1984). C. Panayiotou, J. H. Vera, Polym. J., 16, 89 (1984). A. J. Ashworth, G. J. Price, Thermochim. Acta, 82, 161 (1984). D. Hailemariam, Diploma Paper, TH Leuna-Merseburg, 1985*). M. Krcek, Diploma Paper, TH Leuna-Merseburg , 1986*). C. Booth, G. Gee, M. N. Jones, W. D. Taylor, Polymer, 5, 353 (1964). N.-H. Wang, S. Takashima, H. Masuoka, Kagaku Kogaku Ronbunshu, 15, 313 (1989). N.-H. Wang, S. Takashima, H. Masuoka, Kagaku Kogaku Ronbunshu, 15, 795 (1989). R. Corneliussen, S. A. Rice, H. Yamakawa, J. Chem. Phys., 38, 1768 (1963). C. Booth, C. J. Devoy, Polymer, 12, 320 (1971). C. Booth, C. J. Devoy, Polymer, 12, 309 (1971). Ch. Wohlfarth, unpublished data, TH Leuna-Merseburg, Institute of physical chemistry 1979 - 1993, Martin-Luther-University, Institute of physical chemistry 1993 - 1999*). C. A. Haynes, R. A. Beynon, R. S. King, H. W. Blanch, J. M. Praunsitz, J. Phys. Chem., 93, 5612 (1989). W. R. Moore, R. Shuttleworth, J. Polym. Sci., Pt. A, 1, 1985 (1963).


78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131

241

A. L. Jones, Trans. Faraday Soc., 52, 1408 (1956). E. C. Baughan, A. L. Jones, K. Stewart, Proc. Roy. Soc., London, Ser. A, 225, 478 (1954). G. N. Malcolm, C. E. Baird, G. R. Bruce, K. G. Cheyne, R. W. Kershaw, M. C. Pratt, J. Polym. Sci., Pt. A-2, 7, 1495 (1969). W. R. Moore, R. Shuttleworth, J. Polym. Sci., Pt. A, 1, 733 (1963). K. H. Meyer, E. Wolff, Ch. G. Boissonnas, Helv. Chim. Acta, 23, 430 (1940). S. G. Canagratna, D. Margerison, J. P. Newport, Trans. Faraday Soc., 62, 3058 (1966). H. Ochiai, K. Gekko, H. Yamamura, J. Polym. Sci., Pt. A-2, 9, 1629 (1971). R. C. Osthoff, W. T. Grubb, J. Amer. Chem. Soc., 76, 399 (1954). A. A. Tager, A. I. Suvorova, Yu.S. Bessonov, A. I. Podlesnyak, I.A. Koroleva, L. V. Adamova, Vysokomol. Soedin., Ser. A, 13, 2454 (1971). A. A. Tager, L. V. Adamova, Yu.S. Bessonov, V. N. Kuznetsov, T. A. Plyusnina, V. V. Soldatov, Vysokomol. Soedin., Ser. A, 14, 1991 (1972). T. V. Gatovskaya, V. A. Kargin, A. A. Tager, Zh. Fiz. Khim., 29, 883 (1955). J. Leonard, Van Tam Bui, Polymer, 28, 1041 (1987). A. Hamdouni, J. Leonard, Van Tam Bui, Polymer Commun., 31, 258 (1990). Van Tam Bui, J. Leonard, J. Chem. Soc., Faraday Trans. I, 82, 899 (1986). Van Tam Bui, J. Leonard, J. Chem. Soc., Faraday Trans. I, 81, 1745 (1985). A. H. Abdel-Alim, J. Appl. Polym. Sci., 22, 3597 (1978). U. Messow, Inst. Phys. Chem., Univ. Leipzig, personal communication. F. Cordt, Dissertation, TU Muenchen, 1985. F. Moeller, Dissertation, TU Muenchen, 1989. G. Luengo, G. Rojo, R. G. Rubio, M. G. Prolongo, R. M. Masegosa, Macromolecules, 24, 1315 (1991). N. H. Wang, K. Hattori, S. Takashima, H. Masuoka, Kagaku Kogaku Ronbunshu, 17, 1138 (1991). A. J. Ashworth, G. J. Price, Macromolecules, 19, 362 (1986). A. J. Ashworth, G. J. Price, J. Chem. Soc., Faraday Trans. I, 81, 473 (1985). A. J. Ashworth, G. J. Price, Macromolecules, 19, 358 (1986). I. Noda, Y. Higo, N. Ueno, T. Fujimoto, Macromolecules, 17, 1055 (1984). H. W. Starkweather, Jr., J. Appl. Polym. Sci., 2, 129 (1959). M. Herskowitz, M. Gottlieb, J. Chem. Eng. Data, 30, 233 (1985). Van Tam Bui, J. Leonard, Polym. J., 21, 185 (1989). Y. Iwai, Y. Arai, J. Chem. Eng. Japan, 22, 155 (1989). H. Masuoka, N. Murashige, M. Yorizane, Fluid Phase Equil., 18, 155 (1984). Y. C. Bae, J. J. Shim, D. S. Soane, J. M. Prausnitz, J. Appl. Polym. Sci., 47, 1193 (1993). Ch. Wohlfarth, Plaste & Kautschuk, 40, 272 (1993). G. Luengo, R. G. Rubio, I. C. Sanchez, C. G. Panayiotou, Macromol. Chem. Phys., 195, 1043 (1994). J. Gaube, A. Pfennig, M. Stumpf, J. Chem. Eng. Data, 38, 163 (1993). J. Zhu, Dissertation, Univ. Kaiserslautern, 1991. Y. Iwai, S. Miyamoto, K. Nakano, Y. Arai, J. Chem. Eng. Japan, 23, 508 (1990). Y. Iwai, T. Ishidao, S. Miyamoto, H. Ikeda, Y. Arai, Fluid Phase Equil., 68, 197 (1991). K. Gekko, K. Matsumara, Bull. Chem. Soc. Japan, 46, 1554 (1973). Ch. Wohlfarth, Plaste & Kautschuk, 41, 163 (1994). M. Koester, Diploma Paper, TH Darmstadt, 1994. Ch. Wohlfarth, ELDATA: Int. Electron. J. Phys.-Chem. Data, 1, 113 (1995). C. Grossmann, R. Tintinger, G. Maurer, Fluid Phase Equil., 106, 111 (1995). S. P. V. N. Mikkilineni, D.A. Tree, M. S. High, J. Chem. Eng. Data, 40, 750 (1995). R. B. Gupta, J. M. Prausnitz, J. Chem. Eng. Data, 40, 784 (1995). J. S. Choi, K. Tochigi, K. Kojima, Fluid Phase Equil., 111, 143 (1995). S. Behme, G. Sadowski, W. Arlt, Chem.-Ing.-Techn., 67, 757 (1995). H.-M. Petri, N. Schuld, B. A. Wolf, Macromolecules, 28, 4975 (1995). H.-M. Petri, Dissertation, Univ. Mainz, 1994. Ch. Wohlfarth, ELDATA: Int. Electron. J. Phys.-Chem. Data, 2, 13 (1996). K. Wang, Y. Chen, J. Fu, Y. Hu, Chin. J. Chem. Eng., 1, 65 (1993). D.-Q. Lin, L.-H. Mei, Z.-Q. Zhu, Z.-X. Han, Fluid Phase Equil., 118, 241 (1996). L.R. Ochs, M. Kabri-Badr, H. Cabezas, AIChE-J., 36, 1908 (1990). Ch. Wohlfarth, ELDATA: Int. Electron. J. Phys.-Chem. Data, 2, 163 (1996). K. Wang, Q. Xue, J. Fu, Y. Hu, Huadong Ligong Daxue Xuebao, 22, 330 (1996).

242

132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161

Christian Wohlfarth K. Wang, Q. Xue, Y. Hu, Huadong Ligong Daxue Xuebao, 23, 109 (1997). J. O. Tanbonliong, J. M. Prausnitz, Polymer, 38, 5775 (1997). Ch. Wohlfarth, Macromol. Chem. Phys., 198, 2689 (1997). Ch. Wohlfarth, ELDATA: Int. Electron. J. Phys.-Chem. Data, 3, 47 (1997). K. Tochigi, S. Kurita, M. Ohashi, K. Kojima, Kagaku Kogaku Ronbunshu, 23, 720 (1997). R. K. Surana, R. P. Danner, A. B. De Haan, N. Beckers, Fluid Phase Equil., 139, 361 (1997). H. C. Wong, S. W. Campbell, V. R. Bhetnanabotla, Fluid Phase Equil., 139, 371 (1997). G. Sadowski, L. V. Mokrushina, W. Arlt, Fluid Phase Equil., 139, 391 (1997). C. Mio, K. N. Jayachandran, J. M. Prausnitz, Fluid Phase Equil., 141, 165 (1997). J. S. Aspler, D. G. Gray, Macromolecules, 12, 562 (1979). K. N. Jayachandran, P. R. Chatterji, J. M. Prausnitz, Macromolecules, 31, 2375 (1998). Benczedi, D., Tomka, I., Escher, F., Macromolecules, 31, 3062 (1998). C. Mio, S. Kiritsov, Y. Thio, R. Brafman, J. M. Prausnitz, C. Hawker, E. E. Malmstroem, J. Chem. Eng. Data, 43, 541 (1998). S. Hwang, J. Kim, K.-P. Yoo, J. Chem. Eng. Data, 43, 614 (1998). F. Wie, W. Wenchuan, F. Zhihao, J. Chem. Ind. Eng. China, 49, 217 (1998). J. Kim, K. C. Joung, S. Hwang, W. Huh, C. S. Lee, K.-P. Yoo, Korean J. Chem. Eng., 15, 199 (1998). N. H. Kim, S. J. Kim, Y. S. Won, J. S. Choi, Korean J. Chem. Eng., 15, 141 (1998). Y. Dahong, S. Jibin, L. Xiaohui, H. Ying, J. East China Univ. Sci. Technol., 23, 608 (1997). W. Hiyan, W. Kun, L. Honglai, H. Ying, J. East China Univ. Sci. Technol., 23, 614 (1997). A. Eliassi, H. Modarress, G. A. Mansoori, J. Chem. Eng. Data, 44, 52 (1999). C. Mio, J. M. Prausnitz, Polymer, 39, 6401 (1998). Ch. Wohlfarth, ELDATA: Int. Electron. J. Phys.-Chem. Data, 4, 83 (1998). H.-P. Kany, H. Hasse, G. Maurer, J. Chem. Eng. Data, 44, 230 (1999). J. G. Lieu, M. Liu, J. M. J. Frechet, J. M. Prausnitz, J. Chem. Eng. Data, 44, 613 (1999). J. G. Lieu, J. M. Prausnitz, Polymer, 40, 5865 (1999). J. Kim, E. Choi, K.-P. Yoo, C. S. Lee, Fluid Phase Equil., 161, 283 (1999). L. Ninni, M. S. Camargo, A. J. A. Meirelles, Thermochim. Acta, 328, 169 (1999). K. Wang, Y. Hu, D. T. Wu, J. Chem. Eng. Data, 39, 916 (1994). D.-Q. Lin, Y.-T. Wu, Z.-Q. Zhu, L.-H. Mei, S.-J. Yao, Fluid Phase Equil., 162, 159 (1999). R. N. French, G. J. Koplos, Fluid Phase Equil., 158-160, 879 (1999).

*) all data from the Merseburg group before 1994 have been published in Ref. Ch. Wohlfarth, Vapor-liquid equilibrium data of binary polymer solutions, Physical Science Data 44, Elsevier, Amsterdam, 1994

5

Solubility of Selected Systems and Influence of Solutes 5.1 EXPERIMENTAL METHODS OF EVALUATION AND CALCULATION OF SOLUBILITY PARAMETERS OF POLYMERS AND SOLVENTS. SOLUBILITY PARAMETERS DATA. Valery Yu. Senichev, Vasiliy V. Tereshatov Institute of Technical Chemistry Ural Branch of Russian Academy of Sciences, Perm, Russia

5.1.1 EXPERIMENTAL EVALUATION OF SOLUBILITY PARAMETERS OF LIQUIDS The value of solubility parameter can be calculated from the evaporation enthalpy of liquid at given temperature:1 1/ 2

 ∆Hp − RT   δ =   V  

[5.1.1]

where: ∆Hp V

latent heat of vaporization molar volume

5.1.1.1 Direct methods of evaluation of the evaporation enthalpy For measurement of the evaporation enthalpy of volatile substances, adiabatic apparatuses were developed. They require significant quantities of highly purified substances. The accuracy is determined to a large degree by equipment design and precision of measurement. Most calorimeters that measure a latent heat of vaporization work under isobaric conditions. The measurement of a latent heat of vaporization requires monitoring heat input into calorimeter and the amount of liquid evaporated during measurement time.2-4 In calorimeters of a flowing type,5-6 a liquid evaporates from a separate vessel of a calorimeter. The vapors are directed into the second calorimeter where the thermal capacity of gas is measured. The design of such calorimeters ensures a precise measurement of the stream rate. Heaters and electrical controls permit control of heat flow with high precision and highly sensitive thermocouples measure temperature of gas. There are no excessive

244


thermal losses, thus single-error corrections for the heat exchange can be used to increase precision of measurement. The calorimeters used for the measurement of the heat of reaction can also be used for a measurement of the latent heat of vaporization. These are calorimeters for liquids, micro-calorimeters, mass calorimeters, and double calorimeters.7 The calorimeters with carrier gas are also used.8-9 Evaporation of substance is accelerated by a stream of gas (for example, nitrogen) at reduced pressure. The heat loss by a calorimeter, due to evaporation, is compensated by an electrical current to keep temperature of calorimeter constant and equal to the temperature of the thermostating bath. 5.1.1.2 Indirect methods of evaluation of evaporation enthalpy Because the calorimetric methods of measurement of enthalpy of vapor formation are very difficult, the indirect methods are used, especially for less volatile substances. The application of generalized expression of the first and second laws of thermodynamics to the heterogeneous equilibrium between a condensed phase in isobaric- thermal conditions is given in the Clausius-Clapeyron equation that relates enthalpy of a vapor formation at the vapor pressure, P, and temperature, T. For one component system, the Clausius-Clapeyron equation has the form:7 dP / dT = ∆Hp / T∆V

[5.1.2]

where: ∆V

difference between molar volumes of vapor and liquid

The ratio that neglects volume of a condensed phase with assumption that vapor at low pressure is ideal can be derived from the above equation: d ln P / d (1 / T ) = −∆Hp / R

[5.1.3]

After integration: ln P = −∆Hp / RT + const

[5.1.4]

Introducing compressibility factors of gas and liquids, ∆Z, the Clausius-Clapeyron equation can be written as: d ln P / d (1 / T ) = −∆Hp / R∆Z

[5.1.5]

where: ∆Z

difference between compressibility factors of gas and liquids

The value ∆Z includes corrections for volume of liquid and non-ideality of a vapor phase. The simplifying assumptions give the equation: ln P = A + B / T

[5.1.6]

Approximate dependence of a vapor pressure on inverse temperature is frequently linear but the dependence may also be non-linear because of changing ratio of ∆Hp/∆Z on heating. The mathematical expressions of the dependence lnP on 1/T of real substances in a wide range of temperatures should be taken into account. If ∆Hp/∆Z = a + bT, it results in an equation with three constants:

5.1 Solubility parameters

245

ln P = A + B / T + C lnT

[5.1.7]

In more complicated dependancies, the number of constants may further increase. Another convenient method is based on empirical relation of ∆Hp at 25oC with the normal boiling point, Tb, of non-polar liquids:1 ∆Hp = Tb2 + 23.7Tb − 2950

[5.1.8]

Methods of evaluation of vapor pressure may be divided into static, quasi-static, and kinetic methods. 5.1.1.3 Static and quasi-static methods of evaluation of pair pressure Manometric method10 consists of thermostating with a high precision (0.01K) and vapor pressure measurement by a level of mercury with the help of a cathetometer or membrane zero-manometer. The accuracy of measurement is 0.1-0.2 mm Hg. Ebulliometric method11 is used for a simultaneous measurement of the boiling and condensation temperature that is required for evaluation of purity of a substance and its molecular mass. 5.1.1.4 Kinetic methods These methods were developed based on the molecular kinetic theory of gases. The Langmuir method is based on the evaporation of substance from a free surface into a vacuum. The Knudsen method is based on the evaluation of the outflow rate of a vapor jet from a mesh. The basic expression used in Langmuir method12 is: 1/ 2

 m   2πRT  P =    Stα   M 

[5.1.9]

where: m S t

mass of evaporated substance surface of evaporation time of evaporation.

The Knudsen method13 is based on a measurement of the mass rate of the vapor outflow through a hole. Knudsen proposed the following expression:  ∆m   2πRT    Pk =    S h tβ   M 

1/ 2

[5.1.10]

where: ∆m Sh t β M

mass output of substance surface area of the hole time of vaporization Clausing parameter molecular mass

The method uses special effusion cameras with holes of a definite form, maintaining high vacuum in the system. The method is widely applied to the measurements of a vapor pressure of low volatile substances.

246


The detailed comparative evaluation of experimental techniques and designs of equipment used for determination of enthalpy of evaporation can be found in the appropriate monographs.7,14 Values of solubility parameters of solvents are presented in Subchapter 4.1. 5.1.2 METHODS OF EXPERIMENTAL EVALUATION AND CALCULATION OF SOLUBILITY PARAMETERS OF POLYMERS It is not possible to determine solubility parameters of polymers by direct measurement of evaporation enthalpy. For this reason, all methods are indirect. The underlining principles of these methods are based on the theory of regular solutions that assumes that the best mutual dissolution of substances is observed at the equal values of solubility parameters (see Chapter 4). Various properties of polymer solutions involving interaction of polymer with solvent are studied in a series of solvents having different solubility parameters. A value of a solubility parameter is related to the maximum value of an investigated property and is equated to a solubility parameter of polymer. This subchapter is devoted to the evaluation of one-dimensional solubility parameters. The methods of the evaluation of components of solubility parameters in multi-dimensional approaches are given in the Subchapter 4.1. According to Gee,15 a dependence of an equilibrium swelling of polymers in solvents on their solubility parameters is expressed by a curve with a maximum where the abscissa is equal to the solubility parameter of the polymer. For exact evaluation of δ, a swelling degree is represented by an equation resembling the Gaussian function:

[

Q = Qmax exp −V1 (δ1 − δ 2 )

2

]

[5.1.11]

where: Qmax V1 δ1 , δ 2

the degree of swelling at the maximum on the curve molar volume of solvent solvent and polymer solubility parameters.

Then  1 Q δ 2 = δ1 ±  ln max Q  V1

1/ 2

  

[5.1.12]

The dependence [(1/V1)ln(Qmax/Q)]1/2=f(δ1) is expressed by a direct line intersecting the abscissa at δ1 = δ 2 . This method is used for calculation of the parameters of many crosslinked elastomers.16-19 The Bristow-Watson method is based on the Huggins equation deduced from a refinement of the lattice approach:20 χ = β + (V1 / RT )(δ1 − δ 2 )

2

where: β z m

=(1/z)(1-1/m) a coordination number the chain length.

β may be rewritten as χS entropy contribution to χ (see Chapter 4).

[5.1.13]


247

Accepting that Eq. [5.1.12] represents a valid means of assignment of a constant δ2 to polymer, the rearrangement of this equation gives: δ12 δ2 χ χ  2δ  − =  2  δ1 − 2 − S RT V1  RT  RT V1

[5.1.14]

Now it is assumed that χS is of the order of magnitude suggested above and that, in accordance with the Huggins equation, it is not a function of δ2. Therefore χS/V1 is only about 3% or less of δ 22 /RT for reasonable values of δ 2 of 10-20 (MJ/m3)1/2. Hence Eq. [5.1.14] gives δ2 from the slope and intercept on plot against δ1 (see Figure 5.1.1). This method was improved21 by using calculations that exclude strong deviations of χ. When (χSRT/V1 ≈ const), Eq. [5.1.14] is close to linear (y = A + Bx), where 2 2 Figure 5.1.1. Dependence for equilibrium swelling of y = δ1 − χ S ( RT / V1 ), A = -χS(RT/V1) - δ 2 , crosslinked elastomer on the base of polyether urethane. B = 2δ2, x = δ1. [Adapted, by permission, from V.Yu. Senichev in SynBut δ 2 enters into expression for a tanthesis and properties of cross-linked polymers and compositions on their basis. Russian Academy of Sciences gent of a slope angle and intercept which is cut off on the ordinates axes. This can be Publishing, Sverdlovsk,1990, p.16] eliminated by introduction of a sequential approximation of χS(RT/V1) and grouping of experimental points in areas characterized by a definite interval of values χS(RT/V1). Inside each area χS(RT/V1) →const and Eq. [5.1.14] becomes more precise. The intervals of values χS(RT/V1) are reduced in the course of computations. For n experimental points, the files X (x1, x2,.... xn) and Y (y1, y2, .... yn) are gathered. Tangent of the slope angle is defined by the method of least squares and the current value (at the given stage) of a solubility parameter of a polymer is:

δ2 j =

n

n

n

i =1

i =1

∑ x ∑ y − n ∑ xy i =1

2

n  n  2  ∑ x  − n∑ x  i =1  i =1

[5.1.15]

where: j

a stage of computation

χS(RT/V1) is then calculated using the equation derived from Eqs. [5.1.13] and [5.1.14]:  RT   χ S  = −y i − δ 22 j + 2x i δ 2 j V  1 j

[5.1.16]

248


Table 5.1.1. Modification of δ2 j values during stages of computation

Polymer

j=1

j=2

j=3

j=4

δ2j, (MJ/m3)1/2

polydiene urethane epoxide

17.64

17.88

17.78

17.8

polydiene urethane

17.72

18.17

17.93

17.82

poly(butylene glycol) urethane

19.32

18.89

18.95

18.95

poly(diethylene glycol adipate) urethane

19.42

19.44

19.44

-

where: δ2j

value of δ 2 at the given stage of computation

By sorting of all experimental points into a defined amount of intervals (for 30-50 points it is more convenient to take 5-6 intervals), it is possible to calculate δ 2 for each interval separately. The current average weighted value (contribution of δ 2 is defined), obtained for each interval, is proportional to the amount of points in the interval according to the following formula: δ2 j =

1 k ∑ δ2 k m k M k =1

[5.1.17]

where: k mk M j

= 1,2,.... k, number of intervals number of points in k-interval the total number of points stage of computation.

The shaping of subarrays of points is made in the following order, ensuring that casual points are excluded: 1) account is made in a common array of points of δ 2 and χS(RT/V1)i; 2) partition of a common array into a population of subarrays of χS(RT/V1) in limits defined for elimination of points not included in intervals and points which do not influence consequent stages of computation, 3) reductions of intervals in each of the subarrays (this stage may be repeated in some cases). At the each stage the sequential approximation to constant value of χS(RT/V1) is produced in a separate form, permitting one to take into account the maximum number of points. The procedure gives a seFigure 5.1.2. Dependence of equilibrium swelling of quence of values δ 2 j as shown in Table crosslinked elastomer of polyester urethane (1) and 5.1.1. polybutadiene nitrile rubber (2) on the volume fraction of In still other methods of evaluation,22 acetone in the toluene-acetone mixture. [Adapted, by per- the solvents are selected so that the solubilmission, from V.V. Tereshatov, V.Yu. Senichev, ity parameter of polymer occupies an interA.I. Gemuev, Vysokomol. soed., B32, 412 (1990)]


249

mediate position between solubility parameters of solvents. Assumption is made that the maximum polymer swelling occurs when the solubility parameters of the solvent mixture and polymer are equal. This is the case when the solubility parameter of polymer is lower than the solubility parameter of the primary solvent and the solubility parameter of the secondary solvent is higher. The dependence of the swelling ratio on the composition of solvent mixture has a maximum (see Figure 5.1.2). Such mixed solvents are called symmetric liquids. The reliability of the method is examined by a narrow interval of change of the solubility parameter of a binary solvent. The data obtained by this method in various mixtures differ by no more than 1.5% with the data obtained by other methods. Examples of results are given in Tables 5.1.2, 5.1.3. Table 5.1.2. Values of solubility parameters for crosslinked elastomers from swelling in symmetric liquids. [Adapted, by permission, from V.V. Tereshatov, V.Yu. Senichev, A.I. Gemuev, Vysokomol. soed., B32, 412 (1990).] Elastomer

Symmetric liquids

ϕ H2 at Qmax

δp , (MJ/m3)1/2

Polyether urethane

Toluene( 1) - acetone (2) Cyclohexane (1) -acetone (2)

0.11 0.49

18.4 18.2

Ethylene-propylene rubber

Hexane (1) - benzene (2) Toluene (1) - acetone (2)

0.38 0.32

16.2 18.9

Butadiene-nitrile rubber

Ethyl acetate (1) - acetone (2) Benzene (1) - acetone (1)

0.22 0.27

19.1 19.1

Polyester-urethane

Toluene (1) - acetone (2) o-Xylene (1) - Butanol -1(2)

0.62 0.21

19.6 19.5

Butyl rubber

Octane (1) - benzene (2)

0.26

16.1

Table 5.1.3. Values of solubility parameters of crosslinked elastomers from swelling in individual solvents and symmetric liquids. [Adapted, by permission, from V.V. Tereshatov, V.Yu. Senichev, A.I. Gemuev, Vysokomol. soed., B32, 412 (1990).]

Elastomer

Symmetric liquids

Individual solvents Gee method

Bristow-Watson method

(MJ/m3)1/2 Polyether-urethane

18.3±0.1

17.8, 18.4, 19.4

19.2

16.2

16.1-16.4

-


19.0±0.1

18.9-19.4

18.7

Polyester-urethane

19.5-19.6

19.3, 19.9

19.5

16.1

15.9-16.6

14.9

]

[5.1.18]


Butyl rubber

The calculations were made using equation:

[

δ p = δ mix = δ12 (1 − ϕ′′1 ) + δ 22 ϕ′′1 − ∆Hmix / V12

1/ 2

250


where: ϕ 1′ ∆Hmix V12

volume fraction of a solvent in a binary mixture of solvents causing a maximum of equilibrium swelling experimental value of the mixing enthalpy of components of binary solvent. It can be taken from literature.23-24 the molar volume of binary solvent

If there are no volume changes, V12 can be calculated using the additivity method: V12 = V1 ϕ1 + V2 ϕ 2

[5.1.19]

Attempts25 were made to relate intrinsic viscosity [η] to solubility parameters of mixed solvents. δ 2 of polymer was calculated from the equation: [η] = f(δ1). The authors assumed that the maximum value of [η] is when δ1 = δ 2 of polymer. However, studying [η] for polymethylmethacrylate in fourteen liquids, the authors found a large scatter of experimental points through which they have drawn a curve with a diffusion maximum. Thus the precision of δ 2 values was affected by 10% scatter in experimental data. This method was widely used by Mangaray et al.26-28 The authors have presented [η] as the Gaussian function of (δ1 - δ 2 )2. Therefore, dependence {(1/ V1 ) ln[η] max / [η]}1/ 2 = f ( δ1 ) can be expressed by a straight line intersecting the abscissa at a point for which δ1 = δ 2 . For natural rubber and polyisobutylene, the paraffin solvents and ethers containing alkyl chains of a large molecular mass were studied.26 For polystyrene, aromatic hydrocarbons were used. For polyacrylates and polymethacrylates esters (acetates, propionates, butyrates) were used.27,28 The method was used for determination of δ 2 of many polymers.29-31 In all cases, the authors observed extrema in the dependence of [η] = f(δ1), and the obtained values of δ 2 coincided well with the values determined by other methods. But for some polymers it was not possible to obtain extremum in dependence of [η] = f(δ1).32 A method of the evaluation [η] in one solvent at different temperatures was used for polyisobutylene33 and polyurethanes.34 For polymers soluble in a limited range of solvents, more complex methods utilizing [η] relationship are described.35-37 In addition to the above methods, δ 2 of polymer can be determined from a threshold of sedimentation38 and by critical opalescence.39 In recent years the method of inverse gas-liquid chromatography has been used to evaluate δ 2 of polymers.40,41 One may also use some empirical ratios relating solubility parameters of polymers with some of their physical properties, such as, surface tension42-44 and glass transition temperature.45 The solubility parameters for various polymers are given in Table 5.1.3. Table 5.1.4. Solubility parameters of some polymers36,46

Polymer

δ 3 1/2

3 1/2

(cal/cm )

(MJ/m )

Butyl rubber

7.84

16.0

Cellulose diacetate

10.9

Cellulose dinitrate Polyamide-66

Polymer

δ 3 1/2

3 1/2

(cal/cm )

(MJ/m )

Polydimethylsiloxane

9.53

19.5

22.2

Polydimethylphenyleneoxide

8.6

17.6

10.6

21.6

Polyisobutylene

7.95

16.2

13.6

27.8


9.3

19.0


Polymer

251

δ 3 1/2

(cal/cm )

3 1/2

Polymer

(MJ/m )

δ 3 1/2

(cal/cm )

3 1/2

(MJ/m )

Natural rubber

8.1

16.5

Polymethylacrylate

9.7

19.8

Neoprene

8.85

18.1

Polyoctylmethacrylate

8.4

17.2

Cellulose nitrate

11.5

23.5

Polypropylene

8.1

16.5

Polyacrylonitrile

14.5

29.6

Polypropylene oxide

7.52

15.4

Polybutadiene

8.44

17.2

Polypropylene sulphide

9.6

19.6

Poly-n-butylacrylate

8.7

17.8

Polypropylmethacrylate

8.8

18.0

Polybutylmethacrylate

8.7

17.8

Polystyrene

8.83

18.0

Polybutyl-tert-methacrylate

8.3

16.9

Polyethylene

7.94

16.2

Polyvinylacetate

9.4

19.2

Polyethyleneterephthalate

10.7

21.8

Polyvinylbromide

9.55

19.5

Polyethylmethacrylate

9.1

18.6

Polyvinylidenechloride

12.4

25.3

Polybutadienenitrile (82:18 w)

8.7

17.8

Polyvinylchloride

9.57

19.5

Polybutadienenitrile (75:25)

9.38

19.2

Polyhexyl methacrylate

8.6

17.6


9.64

19.7

Polyglycol terephthalate

10.7

21.8


10.30

21.0

Polydiamylitaconate

8.65

17.7

Polybutadienevinylpyridine (75:25 mas.)

9.35

19.1

Polydibutylitaconate

8.9

18.2

Polybutadienestyrene (96:4)

8.1

16.5

Polysulfone

10.5

21.4

Polybutadienestyrene (87.5:12.5)

8.31

17.0

Polytetrafluorethylene

6.2

12.7


8.5

17.4

Polychloroacrylate

10.1

20.6

Polybutadienestyrene (71.5:28.5)

8.33

17.0

Polycyanoacrylate

14.0

28.6


8.67

17.7

Polyethylacrylate

9.3

19.0

Chlorinated rubber

9.4

19.2

Polyethylenepropylene

7.95

16.2

Ethylcellulose

10.3

21.0

REFERENCES 1 2 3 4 5 6 7 8 9 10

J.H. Hildebrand and R.L. Scott, Solubility of none-electrolytes. 3rd ed., Reinhold, New-York, 1950. J. Mathews, J. Amer. Chem. Soc., 48, 562 (1926). A. Coolidge, J. Amer. Chem. Soc., 52, 1874 (1930). N. Osborn and D. Ginnings, J. Res. Nat. Bur. Stand., 49, 453 (1947). G. Waddington, S. Todd, H. Huffman, J. Amer. Chem. Soc., 69, 22 (1947). J. Hales, J. Cox, E. Lees, Trans. Faraday Soc., 59, 1544 (1963). Y. Lebedev and E. Miroshnichenko, Thermochemistry of vaporization of organic substances, Nauka, Moscow, 1981. F. Coon and F. Daniel, J. Phys. Chem., 37, 1 (1933). J. Hunter, H. Bliss, Ind. Eng. Chem., 36, 945 (1944). J. Lekk, Measurement of pressure in vacuum systems, Mir, Moscow, 1966.

252

11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Valery Yu. Senichev, Vasiliy V. Tereshatov W. Swietoslawski, Ebulliometric measurement, Reinhold, N.-Y., 1945. H. Jones, I. Langmuir, G. Mackay, Phys. Rev., 30, 201 (1927). M. Knudsen, Ann. Phys., 35, 389 (1911). Experimental thermochemistry, Ed. H. Skinner, Intersci. Publ., 1962. G. Gee, Trans. Faraday Soc., 38, 269 (1942). R.F. Boyer, R.S. Spencer, J. Polym. Sci., 3, 97 (1948). R.L. Scott, M. Magat, J. Polym. Sci., 4, 555(1949). P.I. Flory, I. Rehner, J. Chem. Phys., 11, 521 (1943). N.P. Apuhtina, E.G. Erenburg, L.Ya. Rappoport, Vysokomol. soed., A8, 1057 (1966). G.M. Bristow., W.F. Watson, Trans. Faraday Soc., 54, 1731 (1959). V.Yu. Senichev in Synthesis and properties of crosslinked polymers and compositions on their basis. Russian Academy of Sciences Publishing, Sverdlovsk,1990, pp.16-20. V.V.Tereshatov, V.Yu. Senichev, A.I. Gemuev, Vysokomol. soed., B32, 412(1990). V.P. Belousov, A.G. Morachevskiy, Mixing heats of liquids, Chemistry, Leningrad, 1970. V.P. Belousov, A.G. Morachevskiy, Heat properties of non-electrolytes solutions. Handbook, Chemistry, Leningrad, 1981. T. Alfrey, A.I.Goldberg, I.A. Price, J. Colloid. Sci., 5, 251 (1950). D. Mangaray, S.K. Bhatnagar, Rath S.B., Macromol. Chem., 67, 75 (1963). D. Mangarey, S. Patra, S.B. Rath, Macromol. Chem., 67, 84 (1963). D. Mangarey, S. Patra, P.C. Roy, Macromol. Chem., 81, 173 (1965). V.E. Eskin, U. Guravlev, T.N. Nekrasova, Vysokomol. soed., A18, 653 (1976). E. G. Gubanov, S.V. Shulgun, V.Sh. Gurskay, N.A. Palihov, B.M. Zuev, B.E. Ivanov, Vysokomol. soed., A18, 653 (1976). C.I. Scheehan, A.L. Bisio, Rubber Chem. Technol., 39, 149 (1966). A.A. Tager, L.K. Kolmakova, G. Ya. Shemaykina, Ya.S. Vigodskiy, S.N. Salazkin, Vysokomol. soed., B18, 569 (1976). W.R. Song, D.W. Brownawell, Polym. Eng. Sci., 10, 222 (1970). Y.N. Hakimullin, Y.O. Averko-Antonovich, P.A. Kirpichnikov, M.A. Gasnikova Vysokomol. soed., B17, 287 (1975). F.P. Price, S.G. Martin, I.P. Bianchi, J. Polymer Sci., 22, 49 (1956). G.M. Bristow, W.F. Watson, Trans. Faraday Soc., 54, 1731, 1742 (1958). T.G. Fox, Polymer, 3, 11 (1962). K.W. Suh, D.H. Clarke, J. Polymer Sci., A-1, 4, 1671 (1967). V.E. Eskin, A.E. Nesterov, Vysokomol. soed., A8, 1051 (1966). S.K. Ghosh, Macromol. Chem., 143, 181 (1971). A.G. Grozdov, B.N. Stepanov, Vysokomol. soed., B17, 907(1975). A.A. Berlin, V.E. Basin, Fundamentals of adhesion of polymers, Chemistry, Moscow, 1974. S.M. Ygnaytskaya, S.S. Voutskiy, L. Y. Kapluynova, Colloid J., 34, 132 (1972). R.M. Koethen, C.A. Smolders, J. Appl. Polym. Sci., 19, 1163 (1975) R.A. Hayes, J. Appl. Polym. Sci., 5, 318, (1961). E.S. Lipatov, A.E. Nesterov, T.M. Gritsenko, R.A Veselovski, Handbook on polymer chemistry. Naukova dumka, Kiev, 1971. Z.Grubisic-Gallot, M. Picot, Ph. Gramain, H. Benoit, J. Appl. Polym. Sci., 16, 2931(1972). H.-R. Lee, Y.-D. Lee, J. Appl. Polym. Sci, 40, 2087(1990).

5.2 Prediction of solubility parameter

253

5.2 PREDICTION OF SOLUBILITY PARAMETER Nobuyuki Tanaka Department of Biological and Chemical Engineering Gunma University, Kiryu, Japan

5.2.1 SOLUBILITY PARAMETER OF POLYMERS For the purpose of searching for the solvents for a polymer, the solubility parameter of polymers, δ p , is defined as:1-11 δ p = (h 0 / v)

1/ 2

[5.2.1]

where: h0 v

the cohesive enthalpy per molar structural unit for a polymer (cal/mol) the volume per molar structural unit for a polymer (cm3/mol)

because δ p is equivalent to the solubility parameter of solvents, δs, that shows the minimum of the dissolution temperature7 or the maximum of the degree of swelling1,2 for the polymer. For δs, h0 is the molar energy of vaporization that is impossible to measure for polymers decomposed before the vaporization at elevated temperatures, and v is the molar volume of a solvent. The measurements of the dissolution temperature and the degree of swelling are only means to find the most suitable solvents for a polymer by trial and error. In order to obtain easily the exact value of δ p at a temperature, T, the possibility of δ p prediction from the thermal transition behaviors such as the glass transition and the melting has been discussed.9-11 Consequently, it was found that the sum of their transition enthalpies gave h0 in equation [5.2.1] approximately: for crystalline polymers, h0 ≈ hg + hx + hu

T ≤ Tg

[5.2.2]

h0 ≈ hx + hu

Tg < T < Tm

[5.2.3]

T ≤ Tg

[5.2.4]

for amorphous polymers, h0 ≈ hg + hx where: hg hu hx Tg Tm

the glass transition enthalpy per molar structural unit for a polymer the heat of fusion per molar structural unit for a polymer the transition enthalpy per molar structural unit due to ordered parts in the amorphous regions the glass transition temperature; here the onset temperature of heat capacity jump at the glass transition the melting temperature

In the following sections, the physical meanings of hg and hx are shown in the theoretical treatments of the glass transition, and for several polymers, δ p is predicted using these thermodynamic quantities.

254

Nobuyuki Tanaka

5.2.2 GLASS TRANSITION IN POLYMERS The glass transition in polymers is the same kind of physical phenomenon as observed generally for amorphous materials.12 At Tg in the cooling process, polymers are frozen glasses and the molecular motions are restricted strictly. However, the actual states of glasses are dependent on the cooling rate; if the cooling rate is rapid, the glasses formed should be imperfect, such as liquid glasses or glassy liquids.13,14 The annealing for imperfect glasses results in the enthalpy relaxation from imperfect glasses to perfect glasses. At Tg in the heating process, the strong restriction of molecular motions by intermolecular interactions is removed and then the broad jump of heat capacity, Cp, is observed.15 Annealing the glasses, the Cp jump curve becomes to show a peak.15,16 5.2.2.1 Glass transition enthalpy For polymer liquids, the partition function, Ω, normalized per unit volume is given by:10,14,17,18

(

)(2πmkT / h )

Ω = Z N / N!

2

3 Nx / 2

(q / v f )

Nx

{

}

exp −Nxh int / (RT )

[5.2.5]

with vf = qv exp{-hint/(RT)} where: hint h k m N q R vf x Z

the intermolecular cohesive enthalpy per molar structural unit for a polymer Planck’s constant Boltzmann’s constant the mass of a structural unit for a polymer the number of chains the packing factor of structural units for a polymer the gas constant the free volume per molar structural unit for a polymer the degree of polymerization the conformational partition function per a chain

From equation [5.2.5], the enthalpy and the entropy per molar chain for polymer liquids, Hl and Sl, are derived:10 H1 = RT 2 d ln Z / dT + ( 3 / 2)RxT − RxT 2 d ln v f / dT + xh int

[5.2.6]

S1 = (R ln Z + RTd ln Z / dT ) + (3 / 2)Rx − x (R ln v f + RTd ln v f / dT ) + xSd [5.2.7]

with Sd = (3R/2)ln(2πmkT/h2) - (1/x)(R/N)lnN! + Rlnq The first terms on the right hand side of equations [5.2.6] and [5.2.7] are the conformational enthalpy and entropy per molar chain, xhconf and xsconf, respectively.19 Assuming that chains at Tg are in quasi-equilibrium state, the criterions on Tg are obtained: f flow ( = h flow − Tg s flow ) ≈ 0

[5.2.8]

and s flow ≈ 0 (hence hflow ≈ 0)

[5.2.9]

with hflow = Hl/x - 3RTg/2 and sflow = Sl/x - 3R/2 From equations [5.2.8] and [5.2.9], which show the conditions of thermodynamic quasi-equilibrium and freezing for polymer liquids, the conformational enthalpy and entropy per molar structural unit at Tg, hgconf and sgconf, are derived, respectively:

5.2 Prediction of solubility parameter

255

Figure 5.2.2. State models for an amorphous polymer in each temperature range of (a) T≤Tg, (b) TgC=CH-

130

148.5

206.0

-CH=CH-

222

217

243.1

(344)

354

(354.3)

Cyclopentyl

-

676.5

633.0

Cyclohexyl

-

813.5

720.1

Phenyl

735

741.5

683.5

1,4-Phenylene

658

673

704.9

-O-

70

125

115.0

-OH

-

368.5

225.8

-CO-

275

335

263.0

-COO-

310

250

326.6

-COOH

-

318.5

(488.8)

-O-CO-O-

-

375

(441.6)

-CO-O-CO-

-

375

567.3

-CO-NH-

-

600

(443.0)

-O-CO-NH-

-

725

(506.6)

225

225

209.4

-C(CH3)=CH-

-S-

32.0

5.3 Methods of calculation of solubility parameters

Small2

Group

263

Van Krevelen5

Hoy7

F, (cal cm3)1/2 mol-1

-CN

410

480

354.6

-CHCN-

(438)

548.5

(440.6)

-F

(122)

80

41.3

-Cl

270

230

205.1

-Br

340

300

257.9

-I

425

-

-

The Scatchard equation is correct only for nonpolar substances because they have only dispersive interactions between their molecules. Small eliminated from his consideration the substances containing hydroxyl, carboxyl and other groups able to form hydrogen bonds. This method has received its further development due to Fedors’ work,4 who extended the method to polar substances and proposed to represent as an additive sum not only the attraction energy but also the molar volumes. The lists of such constants were published in several works.5-8 The most comprehensive set of contributions to cohesion energy can be found elsewhere.9 Askadskii has shown10 that Fedors’ supposition concerning the additivity of contributions of volume of atoms or groups of atoms is not quite correct because the same atom in an environment of different atoms occupies different volume. In addition, atoms can interact with other atoms in different ways depending on their disposition and this should be taken into account for computation of cohesion energy. Therefore, a new scheme of the solubility parameters calculation was proposed that takes into account the nature of an environment of each atom in a molecule and the type of intermolecular interactions. This approach is similar to that described in the work of Rheineck and Lin.6  ∑ ∆Ε i  δ = i  NA ∑ ∆Vi  i

1/ 2

    

[5.3.3]

where: NA ∆Ei ∆Vi

Avogadro number increment (contribution) to cohesion energy of atom or group of atoms increment to the van der Waals volume of atom

The volume increment ∆Vi of an atom under consideration is calculated as volume of sphere of the atom minus volumes of spherical segments, which are cut off on this sphere by the adjacent covalently-bound atoms: ∆V1 =

where:

4 1 πR 3 − ∑ πh i3 (3R − h i ) 3 i 3

[5.3.4]

264


R hi

hi = R −

van der Waals (intermolecular) radius of a considered atom a height of segment calculated from the formula:

R 2 + d i2 − Ri2 2d i

[5.3.5]

where: di Ri

bond length between two atoms van der Waals radius of the atom adjacent to the covalently-bonded atoms under consideration

The increments to the van der Waals volume for more than 200 atoms in various neighborhoods is available elsewhere.11 Using data from Tables 5.3.2 and 5.3.3, van der Waals volumes of various molecules can be calculated. The increments to the cohesion energy are given in Table 5.3.4. An advantage of this method is that the polymer density that is important for estimation of properties of polymers that have not yet been synthesized does not need to be known. The calculation methods of the solubility parameters for polymers have an advantage over experimental methods that they do not have any prior assumptions regarding interactions of polymer with solvents. The numerous examples of good correlation between calculated and experimental parameters of solubility for various solvents support the assumed additivity of intermolecular interaction energy. The method has further useful development in calculation of components of solubility parameters based on principles of Hansen’s approach.12 It may be expected that useful results will also come from analysis of donor and acceptor parameters used in TDM-approach (see Chapter 4). In Table 5.3.5, the increments required to account for contributions to solubility parameters related to the dipole-dipole interactions and hydrogen bonds are presented.13 Table 5.3.6 contains Hansen’s parameters for some common functional groups. Table 5.3.2. Intermolecular radii of some atoms Atom

R,

Atom

R,

Atom

R,

Atom

R,

C

1.80

F

1.5

Si

2.10

P

1.90

H

1.17

Cl

1.78

Sn

2.10

Pb

2.20

O

1.36

Br

1.95

As

2.00

B

1.65

N

1.57

I

2.21

S

1.8

Table 5.3.3. Lengths of bonds between atoms Bond

di,

Bond

di,

Bond

di,

Bond

di,

C−C

1.54

C−F

1.34

C−S

1.76

N−P

1.65

C−C

1.48

C−F

1.31

C=S

1.56

N−P

1.63

1,40

C−Cl

1.77

H−O

1.08

S−S

2.10

1.34

C−Cl

1.64

H−S

1.33

S−Sn

2.10

C−C

arom

C=C

5.3 Methods of calculation of solubility parameters

265

Bond

di,

Bond

di,

Bond

di,

Bond

di,

C≡C

1.19

C−Br

1.94

H−N

1.08

S−As

2.21

C−H

1.08

C−Br

1.85

H−B

1.08

S=As

2.08

C−O

1.50

C−I

2.21

O−S

1.76

Si−Si

2.32

C−O

1.37

C−I

2.05

O−Si

1.64

P−F

1.55

C=O

1.28

C−P

1.81

O−P

1.61

P−Cl

2.01

C−N

1.40

C−B

1.73

N−O

1.36

P−S

1.81

C−N

1.37

C−Sn

2.15

N−N

1.46

B−B

1.77

C=N

1.31

C−As

1.96

O=N

1.20

Sn−Cl

2.35

C=N

1.27

C−Pb

2.20

O=S

1.44

As−Cl

2.16

1.34

C−Si

1.88

O=P

1.45

As−As

2.42

C−N

arom

C≡N

1.16

C−Si

1.68

N−P

arom

1.58

Table 5.3.4. Values of ∆E*i for various atoms and types of intermolecular interaction required to calculate solubility parameters according to equation [5.3.3] (Adapted from refs. 10,11) Label

∆E*i, cal/mol

C

∆E *C

550.7

H

∆E

* H

47.7

O

∆E

* O

142.6

N

∆E

* N

1205.0

F

Atom and type of intermolecular interaction

∆E

* F

24.2

S

∆E

* S

1750.0

Cl

∆E

* Cl

-222.7

Br

∆E

* Br

583

I

∆E

Double bond

∆E

Dipole-dipole interaction

∆E

Dipole-dipole interaction in nonpolar aprotic solvents of amide type

∆E

Dipole-dipole interaction in nonpolar aprotic solvents as in dimethylsulfoxide

∆E

Aromatic ring

∆E

Hydrogen bond

∆E

Specific interactions in the presence of =CCl2 group

* I

1700

* ≠

-323

* d

1623

* a,N

1623

* a,S * ar * h

∆E *=CCl2

Specific interactions in 3-5 member rings in the presence of O atom

∆E

Isomeric radicals

∆E

* O,r * i

2600 713 3929 2600 2430 -412

266


Table 5.3.5 Increments of atoms or groups of atoms required in equation [5.3.3] 3

∆Vi,

Atoms or their groups

∆Vi,

Atoms or their groups

−CH3

23.2 (22.9)

>CH

11.0 (10.7,10.4)

=CH2

>CH2

3

17.1 (16.8, 16.4)

C

5.0 (4.7,4.5)

21.1

−CH=

15.1 (14.7)

−O−

3.4 (2.7, 2.1)

−OH

10.3 (9.9)

>CO-

18.65 (18.35,18.15)

−NH−

8.8 (8.5)

−NH2

16.1

−CN

25.9

9.0 (8.9)

−Cl

19.9 (19.5)

−F

The values in brackets correspond to one and two neighboring aromatic carbon atoms. In other cases values are given for the aliphatic neighboring carbon atoms.

Table 5.3.6. Hansen’s parameters12

Atom (group)

∆V, cm3/mol

aliphatic

aromatic

−F

18.0

12.55±1.4

~0

~0

−Cl

24.0

12.5±4.2

100±200

100±20

>Cl2

26.0

6.7±1.0

165±10

180±10

−Br

30.0

10.0±0.8

500±10

500±100

−I

31.5

10.3±0.8

1000±200

-

−O

3.8

53±13

1150±300

1250±300

>CO

10.8

36±1

800±250

400±125

>COO

18.0

14±1

1250±150

800±150

−CN

24.0

22±2

500±200

550±200

−NO2

33.5

15±1.5

400±50

400±50

−NH2

19.2

16±5

1350±200

2250±200

>NH

4.5

22±3

750±200

-

−OH

10.0

25±3

4650±400

4650±500

(−OH)n

n 10.0

n (17±2.5)

n (4650±400)

n (4650±400)

−COOH

28.5

8±0.4

2750±250

2250±250

REFERENCES 1 2 3 4

Vδ2h , cal/mol

Vδp , (cal cm/mol)1/2

M. Dunkel, Z. Phys. Chem., A138, 42 (1928). P.A. Small, J. Appl. Chem., 3, 71 (1953). G. Scatchard, J. Amer. Chem. Soc., 56, 995 (1934). R.F. Fedors, Polym. Eng. Sci., 14, 147(1974).

5.4 Mixed solvents - polymer solubility

5 6 7 8 9 10 11 12 13

267

D.W. Van Krevelen, Properties of polymers. Correlations with chemical structure, Elsevier Publishing Comp., Amsterdam-London-New York,1972. A.E. Rheineck and K.F. Lin, J. Paint Technol., 40, 611 (1968). K.L. Hoy, J. Paint Technol., 42, 76 (1970). R.A. Hayes, J. Appl. Polym. Sci., 5, 318 (1961). Y. Lebedev and E. Miroshnichenko, Thermochemistry of evaporation of organic substances. Heats of evaporation, sublimation and pressure of saturated vapor, Nauka, Moscow, 1981. A.A. Askadskii, L.K. Kolmakova, A.A. Tager, et.al., Vysokomol. soed., A19, 1004 (1977). A.A. Askadskii, Yu.I. Matveev, Chemical structure and physical properties of polymers, Chemistry, Moscow, 1983. C. Hansen, A. Beerbower, Solubility parameters. Kirk-Othmer Encyclopedia of chemical technology, 2nd ed., Supplement Vol., 1973, 889-910. H.C. Brown, G.K. Barbaras, H.L. Berneis, W.H. Bonner, R.B. Johannesen, M. Grayson and K.L. Nelson, J. Am. Chem. Soc., 75, 1, (1953).

5.4 MIXED SOLVENTS, A WAY TO CHANGE THE POLYMER SOLUBILITY Ligia Gargallo and Deodato Radic Facultad de Quimica Pontificia Universidad CatÙlica de Chile, Santiago, Chile

5.4.1 INTRODUCTION In general, a mixture is often found to be unexpectedly potent for a particular purpose. Examples are known in several branches of Science and Technology. In the field of solubility, the synergistic effect is, sometimes, spectacular. Innumerable cases of synergism in solvent extraction are known.1-9 Eucaliptus oil has been found to act as a cosolvent, special type of synergistic solvent in case of water-ethanol-gasoline.10 It has been frequently observed that certain polymers can be readily dissolved in mixtures of two or more solvents, whereas they are not soluble in the individual constituents of this mixture. This phenomenon, known as cosolvency is of great practical importance. The scientific and technological importance of polymers has led to extensive study of their solution properties. Most techniques rely on dilute solution methods such as viscometry, light scattering, osmometry, gel swelling or dipole moments. However, these need solvents which completely dissolve the polymer, and many important polymers either are not very soluble, or completely insoluble. In general, polymer solubility normally increases with rising temperature, but negative temperature coefficients are also observed. Increase in polymer molecular weight reduces solubility. Certain combinations of two solvents may become nonsolvents. However, mixtures of non-solvents may sometimes become solvents. In this article, we will review major information for the evaluation of solubility behavior of polymers in binary solvents. Experimental results are compiled and discussed. The emphasis here is on solubility-cosolvency phenomenon, the thermodynamical description of ternary systems and the influence of the polymer structure on preferential adsorption. Finally, interrelationships between polymer structure and thermodynamic properties of the mixture will be discussed relative to the properties and conformation of the dissolved polymer.

268

Ligia Gargallo and Deodato Radic

Some new cosolvent effects will be also described particularly of polymer-supercritical CO2 mixtures. 5.4.2 SOLUBILITY-COSOLVENCY PHENOMENON The addition of a second liquid to a binary liquid-polymer system to produce a ternary system is used widely for a variety of purposes. If the second liquid is a poor solvent, or a precipitant for the polymer, the dissolving potential of the liquid medium can be reduced and eventually phase separation may even occur. This does not necessarily take place in every case and sometimes mixtures of two relatively poor solvents can even produce an enhancement of the solvent power.11-16 The mixed solvent is then said to exhibit a synergistic effect.17 Cosolvency usually refers to a certain range of temperatures of practical use. Within that range, the polymer dissolves in the mixed solvent but not in the pure liquids. The term, “true cosolvency” has been coined to designate those more strict cases, in which the polymer does not dissolve in the pure liquids at any temperature, not just in a given range.18 An analysis of the phase behavior in terms of the free volume theory of polymer solutions has revealed that cosolvency has enthalpic origin.19 In a classic cosolvent system, the single liquids are both very poor solvents for the polymer and the number of polymer-liquid contacts formed in each binary system are not enough to stimulate dissolution of the polymer, except at very short chain lengths. Mixing to produce a ternary system results in a combination of liquid(1)/polymer(3) and liquid(2)/polymer(3) contacts, which according to Cowie et al.20 is sufficient to cause the polymer to dissolve if these contacts are of a different nature, i.e., if the two liquids tend to solvate the polymer at different sites along the chain and so lead to a reinforcement of solvation. If this occurs then, it is also likely that the expanded coil will allow further liquid-polymer contacts to develop even though these may be energetically weak. This idea of a favorable solvation sheath is in agreement with observations made during preferential adsorption studies in ternary systems exhibiting cosolvency.21 Maximum coil expansion is usually found to occur at a mixed solvent composition where there is not preferential adsorption. In other words, the driving force is an attempt to maintain the most favorable composition, thereby minimizing polymer-polymer contacts. However, the balance of interactions giving rise to cosolvency and to inversion in preferential sorption are different, so that both phenomena have to be studied separately. This has been exemplified by a series of systems in which the molecular sizes of the liquid solvents and the nature of their interactions are varied and each plays its role in determining maximum sorption and inversion in preferential sorption.22-25 The interesting phenomenon where a mixture of two poor solvents or nonsolvents for a polymer provides a medium that acts as a good solvent for the polymers26,27 has been the objective of many studies, by light scattering,21,29,30-34 viscometry,35,36 sorption equilibrium,37 and fluorescence.38 From these techniques, it has been possible to appreciate how the second virial coefficient A239,17,40-42 and the intrinsic viscosity [η]13,43-45 preferential adsorption coefficient λ and excimer and monomer emission ratio IE/IM are involved by changing solvent composition. They present ([η], A2) a maximum or a variation at a certain solvent composition where the polymer behaves as through it were dissolved in a good solvent. The quality of solvent or the cosolvent action has been established by determining the magnitude of the miscibility range between the two critical temperatures, UCST and LCST.19,45-58 The application of pressure can widen the miscibility range.59,60


Figure 5.4.1. Phase separation temperature (Tp), as a function of solvent mixture composition, determined on the system PMMA-acetonitrile + chlorobutane, at several polymer concentrations, x(x 102/g cm-3): (a): c = ∆, 12.2; O, 9.17; o 7.14; ∆, 4.36: l, 2.13. (b): c = O, 9.47; £, 8.22; ∆, 5.89; o, 3.69; ∆, 2.17; l, 0.442. From Fernandez-Pierola and Horta.65 (Copyright by Hòthig & Wepf Verlag (Basel), used with permission).

(dTc

/ dx) 0

269

The cosolvency phenomenon was discovered in 1920’s experimentally for cellulose nitrate solution systems.61 Thereafter cosolvency has been observed for numerous polymer/mixed solvent systems. Polystyrene (PS) and polymethylmethacrylate (PMMA) are undoubtedly the most studied polymeric solutes in mixed solvents.62,63 Horta et al.64 have developed a theoretical expression to calculate a coefficient expressing quantitatively the cosolvent power of a mixture (dTc/dx)0, where Tc is the critical temperature of the system and x is the mole fraction of liquid 2 in the solvent mixture, and subscript zero means x→0. This derivative expresses the initial slope of the critical line as a function of solvent composition (Figure 5.4.1).65 Large negative values of (dTc/dx) are the characteristic feature of the powerful cosolvent systems reported.65 The theoretical expression developed for (dTc/dx)0 has been written in terms of the interaction parameters χi for the binary systems:

  V  χ 23 − χ 13 − 2 χ 12  V1   = (−dχ 13 / dT )

[5.4.1]

where: Vi

the volume fraction of polymer and solvent, respectively

All the magnitudes on the right hand side of this equation are to be evaluated at the critical temperature corresponding to x = 0. The expression provides a criterion to predict whether or not the mixed solvent is expected to be a cosolvent of the polymer. When Tc1, is a UCST (as is the case in these phase separation studies), -dχ13/dT > 0 and (dTc/dx)0 has the same sign as the numerator of the equation. Choosing solvent 1 such that Tc2 < Tc1, then (dTc/dx)0 < 0 guarantees that the system will be a cosolvent one. Since χ23 - χ13 > 0, at Tc1, the numerator in the equation [5.4.1] is negative (cosolvent system) if the unfavorable interaction between the two liquids is large enough to compensate for their different affinity towards the polymer. The equation proposed gives a more detailed criterion for cosolvency than the simple criterion of GE > 0. The information needed to predict (dTc/dx)0 from equation [5.4.1] includes the binary interaction parameters of the polymer in each one of the two solvents as a function of temperature, and χ12(GE) for the mixed solvent too. Table 5.4.1 summarizes results reported by Horta et al.64 for some cosolvents of polymethylmethacrylate (PMMA).65

270


Table 5.4.1. Comparison between theory and experiment in cosolvents of PMMA. From Fernandez-Pierola.65 (Copyright by Hòthig & Wepf Verlag (Basel), used with permission) Solvent 1

Solvent 2

Experiment (dTp/dx)0 x 10-2 K

Theory (dTc/dx)0 x 10-2 K

Acetonitrile

Amylacetate

-11.1

-18.0

Acetonitrile

Chlorobutane

-9.4

-9.4

Acetonitrile

CCl4

-12.3

-7.2

Acetonitrile

BuOH

-7. 6

-12.5

4-Chloro-n-butane

Acetonitrile

-3.2

-4.6

1-Chloro-n-butane

BuOH

-17.0

-14.0

Table 5.4.2 Liquid mixtures which are cosolvents of PMMA. From Horta and Fernandez Pierola.64 (Copyright by Butterworth-Heineman Ltd., used with permission) Acetonitrile (AcN) + 4-Heptanone (Hna)a)

Formamide (FA) + Ethanol (EtOH)a)

Acetonitrile (AcN) + Isopentyl acetate (iPac)a)

1-Chlorobutane (BuCl) + Isopentyl alcohol (iPOH)a)

Acetonitrile (AcN) + Pentylacetate (Pac)a)

1-Chlorobutane (BuCl) + Pentyl alcohol (POH)a)

Acetonitrile (AcN) + 1-Chlorobutane (BuCl)a)

1-Chlorobutane (BuCl) + 2-Butanol (sBuOH)b)

Acetonitrile (AcN) + Carbon tetrachloride (CCl4)a)

1-Chlorobutane (BuCl) + Isopropyl alcohol (iPrOH)a)

Acetonitrile (AcN) + Isopentyl alcohol (iPOH)a)

Carbon tetrachloride (CCl4) + 1-Butanol (BuOH)c)

Acetonitrile (AcN) + Pentyl alcohol (POH)a)

Carbon tetrachloride (CCl4) + Ethanol (EtOH)c)

Acetonitrile (AcN) + 1-Butanol (BuOH)a)

Carbon tetrachloride (CCl4) + Methanol (MeOH)c)

Acetonitrile (AcN) + Isopropyl alcohol (iPrOH)a) a) Ref. 65, b) Ref. 47, c) Ref. 6 6

In the case of PMMA, several powerful cosolvent mixtures have been reported.65,46,47 In such systems, a small proportion of liquid 2, added to the polymer-liquid 1 solution, is enough to produce a large decrease in Tp, these cosolvents are accessible to phase separation determinations. In Table 5.4.2, a number of liquid mixtures are listed to act as cosolvents of polymethylmethacrylate (PMMA).65,47,64,67 Intrinsic viscosity [η] has been reported in cosolvent mixtures containing CCl4: CC14/methanol,42,66 ethanol,66 1-propanol,66 1-butanol,66 1-chlorobutane,67 and acetonitrile,68 and acetonitrile/methanol.68 It was also reported the [η] of PMMA in the cosolvent mixture acetonitrile/1-chlorobutane.42 The last system is a powerful cosolvent. Acetonitrile forms powerful cosolvents for PMMA too with other liquids having a wide variety of chemical groups.65 [η] has been reported for PMMA in acetonitrile/pentylacetate as a powerful cosolvent.44 On the contrary, the mixture of 1-chlorobutane/pentyl acetate is a co-nonsolvent of the polymer.44 Mixing acetonitrile, with pentylacetate or mixing acetonitrile with 1-butanol greatly increases solubility and pro-


271

duces a large increase in [η]. This large increase of [η] in the cosolvents contrasts with the approximate constancy of [η] in the l-chlorobutane/pentylacetate mixture. The increase of [η] in a mixed solvent over its weighed mean value in the pure liquids is usually expressed as: ∆[η] = [η] − [η]1 φ1 − [η] 2 φ2

[5.4.2]

where [η]1 and [η]2 refer to values in the pure liquids 1 and 2, and [η] refers to the value in the mixture. This increase in [η] has been attributed to the existence of unfavorable interactions between the two liquids .39 For a given molecular weight, ∆[η] is usually taken to be proportional to the excess in Gibbs function of the mixture: ∆[η](φ) ~ GE(φ)/RT.39 The values of GE and SE for these cosolvent mixtures69,70 at equimolecular composition, and 25°C, are given in Table 5.4.3. Table 5.4.3. Thermodynamic properties of the liquid mixtures used as cosolvents of PMMA. Excess Gibbs function GE, and excess entropy SE, of the binary mixtures at equimolecular composition (at 25°C). From Prolongo et al.44 (Copyright by Butterworth-Heineman Ltd., used with permission) Cosolvent mixture

GE, J mol-1

SE, J mol-1 K-1

MeCN + BuOH

1044

3.70

MeCN + PAc

646

-0.58

MeCN + ClBu

1032

-

The values of GEs have allowed for a qualitative interpretation of the relative values of [η] in these three cosolvent systems studied.44 Mixing cosolvents is much more effective in expanding the polymer coil than increasing temperature.44 In fact, the same increase in [η] experienced by one sample in pure acetonitrile in going from 25°C to 45°C is reached at 25°C by adding just 9% pentyl acetate or 8% 1-butanol or 6% in the case of 1-chlorobutane,71 for polymethylmethacrylate. The sign of ∆[η] was in contradiction with the cononsolvent character attributed to this mixture by cloud point studies.72 This apparent inconsistency could be due to the different range of concentrations in which [η] and cloud point temperature were determined.44 Systematic study of the cosolvency phenomenon has been practically limited to polymethylmethacrylate42,46,64-68,72-81 and polystyrene.45,82,83 The cosolvency is usually explained in terms of the molecular characteristics of the system, specially in terms of molecular interactions. In the powerful cosolvents of PMMA described in the literature72,65,42,44,46,64 one of the liquid components is always either acetonitrile or an alcohol. These are non-random liquids with a certain degree of order in their structure. Two important characteristics seem to be present in these polymer cosolvent systems: the liquid order structure and the tendency of the polymer towards association.65,14,66 The roles of these two factors were considered to interpret solvation of PMMA chains in cosolvent systems.43,14 The mechanisms of cosolvent action have been discussed in terms of the competitive interactions between liquid components and one liquid component and the polymer.44 The best example is the case of acetonitrile and a second liquid having a high proportion of methylene units in its molecule, the unfavorable nitrile-methylene interactions between acetonitrile and PMMA favor

272


the nitrile ester group interaction and an extensive polymer solvation becomes possible. According to Prolongo et a1.,44 the number of methylene units or length of the n-alkyl chain is very important for reaching cosolvency when the second liquid is an ester (acetate). Another factor GE of the acetonitrile + acetate mixtures is larger for long alkyl chains such as in pentyl acetate, favors cosolvency. In the majority of cases, the cosolvent mixtures for PMMA contain either CCl473,77,80,81 or acetonitrile13,42-44,67 as one of the liquid components. A study of the mixture formed by these two liquids and a comparison with the results obtained in the other cosolvents studied before has been also reported.84 The total sorption of the coil (PMMA) was calculated from second virial coefficient and intrinsic viscosity data. According to these authors, acetonitrile can interact favorably with the ester group of PMMA and is unfavorable with its methylene backbone. The role of these opposing interactions and of liquid order in acetonitrile are taken into account to explain the dilute solution properties of PMMA in this cosolvent system.84 On the other hand, in the case of PMMA, in powerful cosolvents, a small proportion of liquid 2, added to the polymer/liquid 1 solution, is enough to produce a large decrease in the phase separation temperature (Tp).64 Katime et al.85 have studied the influence of cosolvency on stereo-complex formation of isotactic and syndiotactic PMMA. The formation of PMMA stereo-complex has been attributed to the interactions between the ester group of the isotactic form and the α-methyl group of the syndiotactic form.86 The stereo-complex was obtained at different compositions of the cosolvent mixtures acetonitrile/carbon tetrachloride, acetonitrile/butyl chloride and butyl chloride/carbon tetrachloride. The results show a high yield of complex formation in pure solvents and when approaches its solvency maximum a decrease of the yield of stereo-complex was observed, indicating that the interactions are impeded.85 The dilute solution viscosity of PMMA in the cosolvent mixture formed by acetonitrile (MeCN) and 1-chloro-n-butane (ClBu) at 25°C has been studied.42,87 The cosolvent effect in this system is extremely large. It has been observed a large increase in the hydrodynamic volume of the macromolecule in solution,42,17 and a step depression in the critical temperature of phase separation (UCST).64,88 The quantitative determination of the magnitude of these effects has been reported.42,64,88 Horta et al.,71 have compared a relative capacities of temperature and of cosolvent mixing on expanding the macromolecular coils and the tendency of the polymer to associate in poor solvents. They have also shown that there is a connection which relates the dependencies of [η] on temperature and solvent composition with the depression in critical temperature (UCST) caused by cosolvency.71 The action of the cosolvent was much more effective in expanding the macromolecule than temperature was.71 These authors have concluded that the comparison between the temperature and solvent composition variations of [η] allows for a correct prediction of the cosolvent depression of the UCST. The comparison between cloud points and [η] - T gives, in general, inconsistent results, but the combination of [η] - T and [η] - φcompensates such inconsistency and establishes a valid link between Tc and [η]. It was found that when the cosolvent power of the binary mixture increases, the complexing capacity decreases. These results were explained by taking into account the excess Gibbs free energy, GE, and the order of the liquid.


273

5.4.3 NEW COSOLVENTS EFFECTS. SOLUBILITY BEHAVIOR Cosolvent effect of alkyl acrylates on the phase behaviour of poly(alkyl acrylate)supercritical CO2 mixtures has been reported.89 Cloud-point data to 220 and 2000 bar are presented for ternary mixtures of poly(butyl acrylate)-CO2-butyl acrylate (BA) and poly(ethylhexyl acrylate)-CO2-ethylhexyl acrylate) (EHA). The addition of either BA or EHA to the respective polymer-solvent mixtures decreases the cloud-point pressures by as much as 1000 bar and changes the pressure-temperature slope of the cloud-point curves from negative to positive, which significantly increases the single-phase region. The literature presents many studies on coil dimensions of synthetic polymers in mixed solvents. Most investigations involve liquid mixtures composed of a good and a poor solvents for the polymer. The action of mixed solvents has been reported to change coil dimensions, not only because of excluded volume effect or due to the interactions existing between the two liquids90,91 but also due to the preferential adsorption of one of the solvent by the polymer. Recently, the behavior of polysiloxanes with amino end-groups in toluene/nitromethane mixtures has been reported.92 This mixture is solvent/non-solvent for the polymer. The transition concentrations separating the concentration domain93 chain flexibility aspects, excluded volume effects94 and total and preferential adsorption coefficients95 of the same system have been discussed. The solubility curves, the cloud point curves and vitrification boundaries for several poly(lactide)-solvent-nonsolvent systems have been reported.96 The liquid-liquid miscibility gap for the systems with the semicrystalline poly(L-lactide) (PLLA) were located in a similar composition range as the corresponding systems with the amorphous poly(DL-lactide) (PDLLA). The solvent-nonsolvent mixtures used for the experiments were: dioxane/water, N-methyl pyrrolidone (NMP)/water, chloroform/methanol and dioxane/methanol. For all PLLA solvent-nonsolvent systems studied solid-liquid demixing was preferred thermodynamically over liquid-liquid demixing. Attempts were made to correlate the experimental finding with predictions on the basis of the Flory-Huggins theory for ternary polymer solutions using interaction parameters derived from independent experiments. Qualitative agreement was found for the relative locations of the liquid-liquid miscibility gaps. The Flory-Huggins description of the solubility curves was less satisfactory. The phase separation processes occurring in poly(L-lactide) (PLLA)-chloroform-methanol mixtures and poly(DL-lactide) (PDLLA)-chloroform-methanol mixtures have been also studied using differential scanning calorimetry, cloud point measurements and optical microscopy.97 It was demonstrated that liquid-liquid demixing occurs in ternary solutions of PDLLA at sufficient high methanol concentrations. For PLLA-containing-solutions, both liquid-liquid demixing processes and soli-liquid demixing processes occur. Only a low cooling rates and high polymer concentration does solid-liquid demixing take place without the interference of liquid-liquid demixing.97 Another interesting effects are the changes of a polyelectrolyte in binary solvents. The complex inter and intramolecular interactions that take place due to the presence of hydrophilic and hydrophobic structural units in the macroion can modify the balance of the interactions and for this reason can change the solubility.98

274


5.4.4 THERMODYNAMICAL DESCRIPTION OF TERNARY SYSTEMS. ASSOCIATION EQUILIBRIA THEORY OF PREFERENTIAL ADSORPTION Polymers dissolved in mixed solvents show the phenomenon of Preferential Adsorption. Experimentally, the preferential adsorption coefficient, λ, is determined. λ is the volume of one of the liquids sorbed in excess by the polymer (per unit mass of polymer). In general, the Flory-Huggins model of polymer solutions is used to describe the Preferential Adsorption. More recently, equation of state theories have been applied.13,43,99-101 Description of experimental results of λ (and another properties as intrinsic viscosities, second virial coefficients, etc.) necessitates the use of correcting terms in the form of ternary interaction parameters. Using equation of state theory it has been shown that such correcting terms can in part be explained by free volume and molecular surface effect.100 Non-random interactions are important in many systems (hydrogen bonding, complex formation, etc.) (see Table 5.4.1). Strongly interacting species can be described taking into account the formation of associates in equilibrium with unassociated molecules (Association Equilibria Theory). Experimental results102 for polymethylmethacrylates in 1,4-dioxane/methanol have been reported , which indicate that the size of the substituent in the polymer ester group exerts an influence on the specific interaction between the methanol molecule and the carbonyl of the ester. In fact, the preferential adsorption of methanol is completely hindered when the lateral group is bulky enough. Similar results have been reported for substituted poly(phenyl methacrylate)s in the mixture tetrahydrofuran/water. The description of these systems was not in agreement with predictions of classical thermodynamic theories.39,103-105 This behavior was analyzed in terms of specific interactions among the components of the ternary system. If the oxygen atoms of 1,4-dioxane can interact specifically with methanol by accepting a proton, then, methanol hydrogen bonds not only to poly(methacrylate)s but also to 1,4-dioxane. The new theoretical formation takes into account the case of solvent which is self-associated and interacts specifically with sites in the polymer chain and with sites in the other solvent molecule. Therefore, it must consider association constants of the solvent molecule and association constants of the self-associated-solvent (2) with (3) the polymer. It was assumed that the polymer molecule has one site for specific interactions with 2, that the constant for such specific interaction between one 2 molecule and one sites is η1 and that the self-association of 2 over the 2 molecule attached to a site in 3 is characterized by constant σ2, η2 or σ2 values, or both, should vary from polymer to polymer depending on the size of the substituent pending from the ester group. In addition, also association constants of 2 with the other solvent molecule (1) have been postulated. These are: η1 and σ1. The number of sites on molecule 1 was called t. The constant for the specific interaction of one 2 molecule with one site in 1 is η1, and the self-association of 2 over the already attached 2 molecule is σ1. With all these constants a quantitative description of the experimental results for the three poly(alkyl methacrylates) (alkyl + Me, Et, iBu) in 1,4-dioxane/methanol was reported.106 In conclusion, the theory applied by Pouchly and Zivny to the simpler case in which one of the liquids is inert, was extended to more complex mixture.106 Table 5.4.4 summarizes the glossary of association constants and interaction parameters used in the theory in the case of poly(alkyl methacrylates). The results that are obtained for the minimum standard deviation in each case are shown on Table 5.4.4.


275

Table 5.4.4. Parameter values giving the minimum deviation (δ) between theory and experiment, for the preferential adsorption coefficient, λ, calculated according to the Association Equilibria Theory. Reprinted with permission from Horta et al.106 (Copyright (1989) American Chemical Society) Polymer

σ

gBA’

ηa

σc

gAC’-r rAgBC’

ηc

δ

1

PMMA

400

0.85

244

400

0.0

5.7

0.034

2

PEMA

400

0.85

244

400

-1.2

59.3

0.028

3

PiBMA

400

0.85

244

400

-2.3

129.5

0.018

4

PMMA

400

0.30

100

400

-3.4

243.2

0.013

5

PEMA

400

0.30

100

400

-2.75

169.2

0.014

6

PiBMA

400

0.30

100

400

-1.6

188.2

0.009

7

PMMA

400

0.85

244

375

-0.1

11.2

0.034

8

PEMA

400

0.85

244

375

-2.2

145.4

0.028

9

PiBMA

400

0.85

244

375

-1.6

100.7

0.035

10

PMMA

400

0.30

100

375

-2.45

195.5

0.013

11

PEMA

400

0.30

100

375

-1.75

125.0

0.016

12

PiBMA

400

0.30

100

375

-0.5

45.6

0.016

The results obtained are a good description of the experimental data on these ternary systems. Effectively, the shape of the variation of λ with solvent composition was well reproduced by the association equilibria theory, as it is shown in Figure 5.4.2. It has extended the same theoretical treatment to other closely related systems, a family of poly(dialkyl itaconates).107 Preferential adsorption coefficient λ was determined and calculated according to the association equilibria theory, and using classical thermodynamic theories.108,103,104,105 The dependence of the preferential adsorption coefficient for poly(dimethyl itaconate) (PDMI), poly(diethyl itaconate) (PDEI), poly(dipropyl itaconate) (PDPI) and poly(dibutyl itaconate) (PDBI), in 1,4-dioxane/methanol mixtures, as a function of the methanol composition (uBO) is shown in Figure 5.4.3. The results are very similar to those of poly(alkyl methacrylates) in the same solvent.106 Comparison between theory and experimental for λ, by using classical thermodynamic theories103-105,108 are shown in Figure 5.4.4. The results obtained are a good description of the experimental data on these ternary systems. Effectively, the shape of the variation of λ with solvent composition was well reproduced by the association equilibria theory. It can be concluded that the association equilibria theory of preferential adsorption in systems with solvent-solvent and solvent-polymer interactions describe in a quantitative way the experimental results of λ. Although, the systems in which the theory has been applied are closely related. More recently, there was no other systems studied to apply this theory.

276


Figure 5.4.2. Comparison of theory and experiment for preferential adsorption coefficient, λ, of poly(alkyl methacrylate)s in 1,4-dioxane-methanol. (uBO=methanol volume fraction). Points: Experimental results from ref.6. (O) PMMA (alkyl = Me); (∆) PEMA (Et); (o) PiBMA (iBu). Association equilibria theory. (8A) Calculated with the parameter values shown in Table 5.4.4 and numbered as 4-6. Curves: (a) PMMA; (b) PEMA; c) PiBMA. (7B) Calculated with the parameter values shown in Table 5.4.4 and numbered as 10-12. Curves: (a) PMMA; (b) PEMA; c) PiBMa. Reprinted with permission from Horta et al.106 (Copyright (1989) American Chemical Society).

Figure 5.4.3. Variation of preferential adsorption coefficient, λ, as function of methanol volume fraction uBO, for PDMI (O), PDEI (∆), PDPI (o), and PDBI (l), at 298 K. Reprinted with permission from Horta et al.107 (Copyright (1990) American Chemical Society).

Figure 5.4.4. Comparison between theory and experiment for the preferential adsorption coefficient, λ, by using classical thermodynamic theories.103 Points: experimental result of λ. Curves: classical thermodynamic theories.108 Reprinted with permission from Horta et al.107 (Copyright (1990) American Chemical Society).


277

5.4.5 POLYMER STRUCTURE OF THE POLYMER DEPENDENCE OF PREFERENTIAL ADSORPTION. POLYMER MOLECULAR WEIGHT AND TACTICITY DEPENDENCE OF PREFERENTIAL ADSORPTION There are some important structural aspects of the polymer which are necessary to take into account in the analysis of the polymer behavior in mixture solvents, such as its polarity, chemical structure, microtacticity, molecular weight. The analysis of these properties shows that they are determinant factors in preferential adsorption phenomena involved. It has been pointed out73 that the effect of tacticity, and particularly the molecular weight, is a complex problem. In the case of poly(2-vinylpyridine), when the polar solvent is preferentially adsorbed, preferential solvation is independent of molecular weight; but when the non-polar solvent is adsorbed, there is a dependence on the molecular weight.62 The reported experimental evidence109-111,67,73,77 seems to show that the coefficient of preferential adsorption λ, for a given polymer in a mixed solvent of fixed composition depends on molecular weight of the polymer sample. It is important to remember, however, that this dependence of λ on M has not been always detected. Particularly for molecular weights lower than a certain value.100,111,91 According to Dondos and Benoit,109 Read,104 and Hertz and Strazielle,112 the dependence of preferential adsorption coefficient λ, with molecular weight or with segment density can be expressed empirically as: λ = λ ∞ + AM −1/ 2

[5.4.4]

λ = λ ∞ + K[η] −1

[5.4.5]

or

where: λ∞ [η] A and K

the value of λ extrapolated to M→ ∞ the intrinsic viscosity constants

It is interesting to note that the variation of λ with M is more pronounced. A in equation [5.4.4] is larger in mixtures which are poor solvents close to θ-conditions than in mixtures with excluded volume.109,111,112 Apparently, there will not be exhaustive results either with the chemical structure of the polymer on the preferential adsorption,113 or the influence of the tacticity on the preferential adsorption. In the last years, investigations regarding the effect of ortho-substituents in polymers with aromatic bulky side groups on the preferential adsorption and viscometric behavior have been reported for poly(phenyl methacrylate) and its dimethyl and diisopropyl ortho derivatives in tetrahydrofuran/water.114 Figure 5.4.5 from ref.114 shows the λvalues for three polymers in THF/water. The λ values diminish when the volume of the side groups increases and there is a strong water adsorption. The behavior reported114 indicates that the cosolvent effect decreases or disappears when the preferential adsorption is very small or is not observed.114 In another publication,102 the systems studied were a series of poly(alkyl methacrylates) including the methyl (PMMA), ethyl (PEMA), isobutyl (PiBMA) and cyclohexyl (PCHMA) substituents in the mixture solvent 1,4-dioxane/methanol. The experimental

278


results113 indicate that the size of the substituents on the polymer ester group exerts an influence on the specific interaction between the methanol molecule and the carbonyl of the ester. It has been shown that the preferential adsorption of methanol is completely hindered when the lateral groups are bulky enough. Several attempts have been made to take into account the geometrical character of the polymer segment and the solvent molecule. None of them seems to give a Figure 5.4.5. Variation of the preferential adsorption as a unique explanation for the experimental refunction of solvent composition (∆) PPh (Gargallo et al., 1984): (l) PDMPh: (O) PDPPh. From Gargallo et al.114 sults about the cosolvent effect. The most important task in the field of (Copyright by Springer Verlag, used with permission). research on ternary systems polymer in binary solvents is to examine the state of the macromolecular chain in solution and the analysis of the changes in solubility and then in composition involved in the total system itself. The goal of this review was to present aspects of the preferential adsorption phenomena of solvents and polymers with a focus on their thermodynamic aspects. The idea behind this was to attract the attention of polymeric physico-chemists to this area, which is sufficiently related to a lot of different effects. In fact, preferential adsorption occupies a special place in the solubility of polymers in mixed solvents. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Y. A. Zolotov, and L. G. Gamilova, J. Inorg. Nucl. Chem., 31, 3613 (1969). V. Pandu Ranga Rao, and P.V.R. Bhaskara Sarma, J. Inorg. Nucl. Chem., 31, 2151 (1969). T.V. Healy, J. Inorg. Chem., 31, 49 (1969). M.A. Carey and C.V. Banks, J. Inorg. Nucl. Chem., 31, 533 (1969). L. Newman and P. Klotz, Inorg. Chem., 5, 461 (1966). H.M.N.H. Irving and N.S. Al-Niaimi, J. Inorg. Nucl. Chem., 27, 1671, 2231 (1965). H.M.N.H. Irving and N.S. Al-Niaimi, J. Inorg. Nucl. Chem., 21, 1671, 2231 (1965). L. Schaefer, Am. J. Pharm., 85, 439 (1913). Y. Marcus, Chem. Rev., 63, 161 (1963). A.F.M. Barton and J. Tjandra, Fuerl, 68, 11 (1989). L. Gargallo, D. Radic’, and E. Zeidan, Bol. Soc. Chil. Qufm., 27, 15 (1982). A.A. Abdel-Azim, S.S. Moustafa, M.M. Edl Dessouky, F. Abdel-Rehim, and S.A. Hassan, Polymer, 27, 15 (1982). R.M. Masegosa, M. G. Prolongo, I. Hernandez-Fuentes, A. Horta, Macromolecules, 17, 1181 (1984). J.R Ochoa, B. Caballero, R. Valenciano and I. Katime, Mater. Chem. Phys., 9, 477 (1983). J.M.G. Cowie and J.T. McCrindle, Eur. Polym. J., 8, 1325 (1972). T. Spychaj. Angew, Makromol. Chem., 149, 127 (1987). P. Munk, M.T. Abijaoude, M.E. Halbrook, J. Polym. Sci. Polym. Phys., 16, 105 (1978). a) S. Matsuda and K. Kamide. Polym, J., 19, 2, 211 (1987); b) S. Matsuda and K. Kamide, Polym. J., 19, 2, 203 (1987); c) B.A. Wolf, and R.J. Molinari, Makromol. Chem., 173, 241 (1973). C. Strazielle and H. Benoit, J. Chem. Phys., 38, 675 (1961). J.M.G. Cowie, J. Polym. Sci., C23, 267 (1968). L. Gargallo, D. Radicand, I. Femandez-Pierola, Makromol. Chem. Rapid. Commun., 3, 409 (1982). C. Feyreisen, M. Morcellet and C. Loucheux, Macromolecules, 11, 620 (1978). B. Carton, V. Boltigliori, M. Morcellet, and C. Loucheux, Makromol. Chem., 179, 2931 (1978). R. Noel, D. Patterson and T. Somcynsky, J. Polym. Sci., B42, 561 (1960).


25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79

279

R. Gavara, B. Celda, A. Campos, Eur. Polym. J., 22, 373 (1986). D.K. Sarkar, S.R Palit, J. Polym. Sci., C30, 69 (1970). K. Ishikawa, T. Kawai, J. Chem. Soc. Jpn. Ind. Chem. Sci., 55, 173 (1952). H. Yamakawa, J. Chem. Phys., 46, 3, 973 (1967). K. Takashima, K. Nake, M. Shibata and H. Yamakawa, Macromolecules, 7, 641 (1974). P. Kratochvil, J. Polym. Sci. Polym. Symp., 50, 487 (1975); 55, 173 (1952). J.M.G. Cowie, Pure App1. Chem., 23, 355 (1970). A. Campos, B. Celda, J. Mora and J.E. Figueruelo, Polymer, 25, 1479 (1984). W. Wunderlich, Makromol. Chem., 182, 2465 (1981). M. Nakata andN. Numasawa, Macromolecules, 18, 1736 (1985). B. Friederich and K. Prochazka, Eur. Polym. J., 15, 873 (1979). I. Katime, P.M. Sasiaand B. Eguia, Eur. Polym. J., 24, 12, 1159 (1988). J. Pouchly and A. Zivny, Makromol. Chem., 186, 37 (1985). R.M. Masegosa, I. Hernandez-Fuentes, I. Fernandez Pierola, A. Horta, Polymer, 28, 231 (1987). A. Dondos and D. Patterson, J. Polym. Sci., A-2, 7, 209 (1969). C. Tsitsilianis, E. Pierri and A. Dondos, J. Polym. Sci. Polym. Lett. Ed., 21, 685 (1983). M. Nakata, Y. Nakano and K. Kawata, Macromolecules, 21, 2509 (1988). M. G. Prolongo, R.M. Masegosa, I. Hernandez-Fuentes and A. Horta, Macromolecules, 14, 1526 (1981). A. Horta and I. Fernandez-Pierola, Macromolecules, 14, 1519 (1981). M.G. Prolongo, R.M. Masegosa, I. Hernandez-Fuentes and A. Horta, Polymer, 25, 1307 (1984). F. Viras and K. Viral, J. Polym. Sci. Phys. Ed., 26, 2525 (1988). J.M.G. Cowie and IT McEwen, Macromolecules, 7, 291 (1974). B.A. Wolf and G.J. Blaum, Polym. Sci., 13, 1115 (1975). J.M.G. Cowie and IT McEwen, Polymer, 24, 1449 (1983). J.M.G. Cowie and IT McEwen, Polymer, 24, 1445 (1983). J.M.G. Cowie and IT McEwen, Polymer, 24, 1453 (1983). B.A. Wolf, J.W. Breitenbach and H. Senftl, J. Polym. Sci., C31, 345 (1970). J.M.G. Cowie, IT McEwen and M.T. Garay, Polymer Commun., 27, 122 (1986). J.M.G. Cowie and IT McEwen, J. Polym. Sci. B. Phys. Ed., 25, 1501 (1987). J.M.G. Cowie, M.A., Mohsin, I.J. McEwen, Polymer, 28, 1569 (1987). J.W. Breitenbach, B.A. Wolf, Makromol.Chem., 117, 163 (1968). B.A. Wolf, J.W. Breitenbach and H. Senftl, Monatsh, Chem., 101, 57 (1970). B.A. Wolf, J.W. Breitenbach and J.K. Rigler, Ang. Makromol. Chem., 34, 177 (1973). J.M.G. Cowie, IT McEwen, J. Polym. Sci. Phys. Ed., 25, 1501 (1987). G. Blaum and B.A. Wolf, Macromolecules, 9, 579 (1976). B.A. Wolf and G. Blaum, Makromol. Chem., 177, 1073 (1976). F.D. Miles, Cellulose Nitrate, Oliver and Boyd Ed., Imperial Chem. Industries Ltda., London, 1955, Chapter V. L. Gargallo and D. Radic’, Adv. Colloid. Interface Sci., 21, 1-53 (1984). A. Dondos and D. Patterson, J. Polym. Sci., A-2, 5, 230 (1967). A. Horta and I. Fernandez-Pierola, Polymer, 22, 783 (1981). I. Fernandez-Pierola and A. Horta, Makromol. Chem., 182, 1705 (1981). P. Ch. Deb and S.R Palit, Makromol. Chem., 166, 227 (1973). I. Katime, J.R Ochoa, L.C. Cesteros and J. Penafiel, J. Polymer Bull., 6, 429 (1982). Z. Wenging and X. Chengwei, Fenzi Kexu Xuebao, 2, 73 (1982). R.M. Masegosa, M.G. Prolongo, I. Hernandez-Fuentes and A. Horta, Ber. Bunseges. Phys. Chem., 58, 103 (1984). I. Fernandez-Pidrola and A. Horta, J. Chim. Phys., 77, 271 (1980). A. Horta, M.G. Prolongo, R.M. Masegosa and I. Hernandez-Fuentes, Polymer Commun., 22, 1147 (1981). I. Fernandez-Pierola and A. Horta, Polym. Bull., 3, 273 (1980). I. Katime and C. Strazielle, Makromol. Chem., 178, 2295 (1977). I. Katime, R. Valenciano and M. Otaduy, An. Quim., 77, 405 (1981). I. Katime, M.B. Huglin and P. Sasia, Eur. Polym. J., 24, 561 (1988). M.D. Blanco, J.M. Teij6n, J.A. Anrubia and I. Katime, Polym. Bull., 20, 563 (1988). I. Katime, J. Tamarit and J.M. Teijon, An. Quim., 75, 7 (1979). I. Fernandez-Pierola, and A. Horta, IUPAC Symposium Macromolecules, 1980, pp. 3, 47. I. Fernandez-Pierola, Tesis Doctoral, Universidad Complutense, Madrid, 1981.

280

80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114


M.G. Prolongo, R.M. Masegosa, L. de Blas, I. Hernandez-Fuentes, and A. Horta, Commun. to the IUPAC-Macro’83, Bucarest, September, 1983. N. Vidyarthi, M. Gupta and S.E. Palit, J. Indian Chem. Soc., 61, 697 (1979). P.C. Read, and S.R. Palit, Makromol. Chem., 128, 123 (1972). J.M.G. Cowie and J.T. McGrindle, Eur. Polym. J., 8, 1185 (1972). J. Vasquez, L. de Blas, M. G. Prolongo, R.M. Masegosa, I. Hernandez-Fuentes, and A. Horta, Makromol. Chem., 185, 797 (1984). I. Katime, J.R Quintana, J. Veguillas, Polymer, 24, 903 (1983). F. Bosscher, D. Keekstra and G. Challa, Polymer, 22, 124 (1981). M.G. Prolongo, R.M. Masegosa, I. Hernandez-Fuentes and A. Horta, IUPAC on Macromolecules, 1980, pp. 3, 51. I. Fernandez-Pierola and A. Horta, Symp. on Macromolecules, Italy, 1980, pp. 3, 47. M.A. McHugh, F. Rindfleisch, P.T. Kuntz, C. Schmaltz, M. Buback, Polymer, 39 (24), 6049 (1998). A. Dondos, P. Rempp and H. Benoit, J. Polym. Sci., C30, 9 (1970). A. Dondos and H. Benoit, Macromolecules, 4, 279 (1971). G. Grigorescu, S. Ioan, V. Harabagin and B.C. Simionescu, Eur. Polym. J., 34, 827-832 (1998). G. Grigorescu, S. Ioan, V. Harabagin and B.C. Simionescu, Rev. Roum. Chim., 42, 701 (1997). G. Grigorescu, S. Ioan, V. Harabagin and B.C. Simionescu, Macromol. Rep., A33, 163 (1966). G. Grigorescu, S. Ioan, C. Cojocaru and B.C. Simionescu, Pure Appl. Chem. A34, 533 (1997). P. van de Witte, P.J. Dijkstra, J.W.A. van den Berg and J. Feijen, J. Polym. Sci., Part B: Polym. Phys., 34, 2553 (1996). P. van de Witte, A. Borsma, H. Esselbrugge, P.J. Dijkstra, J.W.A. van den Berg and J. Feijen, Macromolecules, 29, 212 (1996). V.O. Aseyev, S.I. Klenin and H. Tenhu, J. Polym. Sci. Part B: Polym. Phys., 36, 1107 (1998). A. Horta, Macromolecules, 12, 785 (1979). A. Horta, Macromolecules, 18, 2498 (1985). A. Horta, and M. Criado-Sancho, Polymer, 23, 1005 (1982). I. Katime, L. Gargallo, D. Radic’ and A. Horta, Makromol. Chem., 186, 2125 (1985). T. Kawai, Bull. Chem. Soc. Jpn., 26, 6 (1953). B.E. Read, Trans. Faraday Soc., 56, 382 (1960). W.R. Krigbaum and D.K. Carpenter, J. Phys. Chem., 59, 1166 (1955). A. Horta, D. Radic’ and L. Gargallo, Macromolecules, 22, 4267 (1989). A. Horta, L. Gargallo and D. Radic’, Macromolecules, 23, 5320 (1990). P.W. Morgan and S.L. Kwolek, Macromolecules, 8, 104 (1975). A. Dondos and H. Benoit, Makromol. Chem., 133, 119 (1970). E.F. Casassa, Polymer J., 3, 517 (1972). A. Dondos, and H. Benoit, in Order in Polymer Solutions, K. Solc. Ed., Gordon and Breach, London, 1976. M. Hertz and C. Strazielle, in Order in Polymer Solutions, K. Solc Ed., Gordon and Breach, London, 1976, pp 195. M. Yamamoto and J.L. White, Macromolecules, 5, 58 (1972). L. Gargallo, N. Hamidi, I. Katime and D. Radic, Polym. Bull., 14, 393 (1985).

5.5 The phenomenological theory of solvent effects

281

5.5 THE PHENOMENOLOGICAL THEORY OF SOLVENT EFFECTS IN MIXED SOLVENT SYSTEMS Kenneth A. Connors School of Pharmacy, University of Wisconsin, Madison, USA

5.5.1 INTRODUCTION We do not lack theories dealing with solvent effects on chemical and physical processes, as is made clear by other sections and authors in the present book. Some of these theories are fundamental in the sense that they invoke detailed physical descriptions of molecular phenomena (electrostatic interactions or the dispersion interaction, for example) whereas others are extensively empirical (such as the UNIQUAC and UNIFAC schemes for estimating activity coefficients, or extrathermodynamic correlations with model processes, exemplified by the Dimroth-Reichardt ET value). Given the abundance of theoretical and empirical approaches, it might seem that new attacks on the general problem of solvent effects would be superfluous. Yet when a solvent effect problem (in particular a solubility problem) arises, the extant theories often are in some measure inadequate. The empirical approaches tend to constitute special rather than general methods of attack, and the physical theories are either too complexly detailed or must be overly simplified to be usable. (It should be noted that our present concern is dominated by an interest in pharmaceutical systems, and therefore by aqueous and mixed aqueous solvents). The consequence is that a chemist or pharmacist confronted with a solubility problem (and such problems usually arise in situations allowing little commitment of time to the problem) often finds it more fruitful to solve the problem experimentally rather than theoretically. This is perfectly valid, but seldom does the solution lead to deeper understanding, and moreover the time required for the experimental effort may be excessive. Another factor is the frequent availability of only milligram quantities of material. It is these considerations that led to the development of this phenomenological theory of solvent effects in mixed aqueous-organic solvent systems. The theory is termed “phenomenological” because it includes elements of description such as equilibrium constants whose evaluation is carried out experimentally, yet it is based on physicochemical ideas. 5.5.2 THEORY In the following development, the symbol x represents mole fraction, c is the molar concentration, component 1 is water, component 2 is a water-miscible organic cosolvent, and component 3 is the solute. 5.5.2.1 Principle Before we can carry out any solution chemistry we must have a solution, and so we begin with the process of dissolution of a solid solute in a liquid solvent (which may itself be a mixture), the system being at equilibrium at constant temperature T.1 The experimentally measured equilibrium solubility of the solute is x3. Eq. [5.5.1] gives the free energy of solution per molecule, where k is the Boltzmann constant. ∆G *soln = kT ln x 3

[5.5.1]

282

Kenneth A. Connors

Conventionally the standard free energy of solution is given by eq. [5.5.2], 0 ∆Gsoln = kT ln a 3

[5.5.2]

where a3 is the mole fraction solute activity. Since we can write a3 = f3x3, eq. [5.5.3] relates ∆G*soln and ∆G0soln . 0 + kT ln f 3 ∆G*soln = ∆Gsoln

[5.5.3]

The traditional approach would be to develop a theory for the activity coefficient f3, which measures the extent of nonideal behavior. This seems to be a very indirect manner in which to proceed, so we will henceforth make no use of activity coefficients, but instead will develop an explicit model for ∆G*soln . The general principle is to treat ∆G*soln as the sum of contributions from the three types of pairwise interactions: solvent-solvent interactions, which give rise to the general medium effect; solvent-solute interactions, or the solvation effect; and solute-solute interactions (the intersolute effect in the present context). Thus we write. ∆G*soln = ∆G gen med + ∆Gsolv + ∆G intersol

[5.5.4]

Our problem is to develop explicit expressions for the solvent-dependent quantities on the right-hand side of eq. [5.5.4]. 5.5.2.2 The intersolute effect: solute-solute interactions There are two contributions to the intersolute effect. One of these comes from solute-solute interactions in the pure solute, which for solid solutes constitutes the crystal lattice energy. We will make the assumption that this contribution is independent of the nature (identity and composition) of the solvent. Usually this is a valid assumption, but exceptions are known in which the composition of the solid depends upon the composition of the solvent. Theophylline, for example, forms a hydrate in water-rich solvents, but exists as the anhydrous form in water-poor solvents; thus its crystal energy varies with the solvent composition.2-4 Although the final theory is capable of empirically describing the solubility of such systems, it lacks a valid physical interpretation in these cases. Fortunately such solid solute behavior is not common, and our assumption that the pure solute interaction energy is solvent-independent is usually a good one. The second contribution to the intersolute effect comes from solute-solute interactions in the solution phase. In pharmaceutics our motivation for incorporating organic solvents into an aqueous system commonly arises from an unacceptably low equilibrium solubility of a drug in pure water. This means that in water and in water-rich mixed solvents the extent of solution phase solute-solute interactions will be negligible because the solute concentration is in the extremely dilute range. At higher concentrations of the organic cosolvent it is true that the solute concentration may rise well above the dilute range, but in some degree this is offset by the diminished tendency for solute-solute interaction in such systems. Thus the hydrophobic interaction is sharply decreased by incorporating organic cosolvents. We will recognize these solution phase solute-solute interactions as a possible source of perturbation in our theory because of our assumption either that they are negligible, or that they do not vary with solvent composition.


283

5.5.2.3 The solvation effect: solute-solvent interaction Our approach is to treat solvation as a stoichiometric equilibrium process. Let W symbolize water, M an organic cosolvent, and R the solute. Then we postulate the 2-step (3-state) system shown below. RW 2

+ M

RWM

+ M

K1

K2

RWM RM2

[5.5.5]

+ W

+ W

[5.5.6]

In this scheme K1 and K2 are dimensionless solvation equilibrium constants, the concentrations of water and cosolvent being expressed in mole fractions. The symbols RW2, RWM, RM2 are not meant to imply that exactly two solvent molecules are associated with each solute molecule; rather RW2 represents the fully hydrated species, RM2 the fully cosolvated species, and RWM represents species including both water and cosolvent in the solvation shell. This description obviously could be extended, but experience has shown that a 3-state model is usually adequate, probably because the mixed solvate RWM cannot be algebraically (that is, functionally) differentiated into sub-states with data of ordinary precision. Now we further postulate that the solvation free energy is a weighted average of contributions by the various states, or ∆Gsolv = ∆GWW FWW + ∆GWM FWM + ∆G MM FMM

[5.5.7]

where FWW, FWM, and FMM are fractions of solute in the RW2, RWM, and RM2 forms, respectively. Eq. [5.5.7] can be written ∆Gsolv = (∆GWM − ∆GWW )FWM + (∆G MM − ∆GWW )FMM + ∆GWW

[5.5.8]

By combining definitions of K1, K2, FWM, and FMM we get FWM =

K1 x 1 x 2 x 12 + K1 x 1 x 2 + K1K 2 x 22

FMM =

K1K 2 x 22 x 12 + K1 x 1 x 2 + K1K 2 x 22

[5.5.9]

for use in eq. [5.5.8] Now observe this thermodynamic cycle [5.5.10]:

From this cycle we get ∆G1 = ∆GWM − ∆GWW = −kT ln K1

[5.5.11]

284

Kenneth A. Connors

A similar cycle yields eq. [5.5.12]. ∆G 2 = ∆G MM − ∆GWW = −kT ln K1K 2

[5.5.12]

Combination of eqs. [5.5.8] - [5.5.12] then gives ∆Gsolv =

(−kT ln K1 )K1 x 1 x 2 + (−kT ln K1K 2 )K1K 2 x 22 x 12 + K1 x 1 x 2 + K1K 2 x 22

+ ∆Gww

[5.5.13]

Obviously when x2 = 0, ∆Gsolv = ∆GWW.5 Eq. [5.5.13] is the desired expression relating the solvation energy to the solvent composition. 5.5.2.4 The general medium effect: solvent-solvent interactions Here we make use of Uhlig’s model,6 writing eq. [5.5.14] as the energy required to create a molecular-sized cavity in the solvent. ∆G gen med = gAγ

[5.5.14]

In eq. [5.5.14] g is a curvature correction factor, an empirical quantity that corrects the conventional surface tension γ for the curvature of the cavity needed to contain a solute molecule. A is the surface area of this cavity in 2 molecule-1; in Sections 5.5.3 and 5.5.4 we treat the meaning of A in more detail, but here we only make the assumption that it is essentially constant, that is, independent of x2. There is a subtlety in assigning the value of γ, for implicit in our model (but treated more fully in reference 1) is the condition that the γ of eq. [5.5.14] is the surface tension of the cavity surface at its equilibrium composition. But this is the composition of the solvation shell immediately adjacent to the molecule, and this is in general different from the composition (x1, x2) of the bulk solvent mixture. Let f1 and f2 be the equilibrium mean fractional concentrations of water and cosolvent, respectively, in the solvation shell, so f1 + f2 = 1. These fractions are defined f1 =

1 (2FWW + FWM ) 2

[5.5.15]

f2 =

1 (FWM + 2FMM ) 2

[5.5.16]

Now we define, for use in eq. [5.5.14], γ = γ 1f1 + γ 2 f 2

[5.5.17]

γ = γ 1 + (γ 2 − γ 1 )f 2

[5.5.18]

where γ1 and γ2 are the surface tensions of pure component 1 (water) and 2, respectively. Combining eqs. [5.5.14, 5.5.16, and 5.5.18] gives for the general medium effect ∆G gen med = gAγ 1 +

gAγ ′K1 x 1 x 2 + 2gAγ ′K1K 2 x 22 x 12 + K1 x 1 x 2 + K1K 2 x 22

[5.5.19]


285

where γ ′ = ( γ 2 − γ 1 ) / 2. Notice that the general medium and solvation effects are coupled through the solvation constants K1 and K2. When x2 = 0, eq. [5.5.19] yields ∆Ggen med = gAγ1. We interpret this as a quantitative expression for the hydrophobic effect. In general, eq. [5.5.19] describes the solvophobic effect. This is a phenomenological description, not a detailed structural description. 5.5.2.5 The total solvent effect The solution free energy is now obtained by inserting eqs. [5.5.13] and [5.5.19] into eq. [5.5.4]. We obtain ∆G*soln ( x 2 ) = gAγ 1 + ∆G intersol + ∆GWW +

(gAγ′ − kT ln K1 )K1 x 1 x 2 + (2gAγ′ − kT lnK1K 2 )K1K 2 x 22 x 12 + K1 x 1 x 2 + K1K 2 x 22

[5.5.20]

When x2 = 0, eq. [5.5.20] gives ∆G*soln (x 2 = 0) = gAγ 1 + ∆G intersol + ∆GWW

[5.5.21]

With the Leffler-Grunwald delta operator symbolism7 we define δ M ∆G * = ∆G*soln (x 2 ) − ∆G*soln (x 2 = 0)

[5.5.22]

which, applied to eqs. [5.5.20] and [5.5.21], gives our final result: * δ M ∆Gso ln =

(gAγ′ − kT ln K1 )K1 x 1 x 2 + (2gAγ′ − kT ln K1K 2 )K1K 2 x 22 x 12 + K1 x 1 x 2 + K1K 2 x 22

[5.5.23]

The quantity δ M ∆G*soln can be read “the solvent effect on the solution free energy.” Because of eq. [5.5.1], δ M ∆G* is proportional to the “relative solubility,” log[( x 3 ) x2 / ( x 3 ) x2 = 0 ], that is, the logarithm of the solubility in the mixed solvent of composition x2 relative to the solubility in pure water. The subtraction that yields eq. [5.5.23], a workable equation with just three unknown parameters (gA, K1, and K2), has also prevented us from dealing with absolute solubilities. 5.5.3 APPLICATIONS 5.5.3.1 Solubility It will be no surprise that the first use of eq. [5.5.23] was to describe the equilibrium solubility of solid nonelectrolytes in mixed aqueous-organic solvents.1 Equilibrium solubility in mol L-1, c3, is converted to mole fraction, x3, with eq. [5.5.24], where ρ is the saturated solution density, w is the wt/wt percentage of organic cosolvent, and M1, M2, M3 are the molecular weights of water, cosolvent, and solute.8 x3 =

c3

 w (1 − w)   c 3 (1000ρ − c 3 M 3 ) M + M  2 1  

[5.5.24]

286

Figure 5.5.1. Solvent effect on the solubility of diphenylhydantoin. Cosolvents, top to bottom: glycerol, methanol, ethanol. The smooth lines were drawn with eq. 5.5.23. (Reproduced with permission from the Journal of Pharmaceutical Sciences reference 1.)

Kenneth A. Connors

Figure 5.5.2. Solvent effect on the solubilities of barbituric acid derivatives in ethanol-water mixtures. Top to bottom: metharbital, butabarbital, amobarbital. The smooth lines were drawn with eq. 5.5.23. (Reproduced with permission from the Journal of Pharmaceutical Sciences, reference 1.)

The free energy of solution per molecule is then calculated with eq. [5.5.1], δ M ∆G* is found with eq. [5.5.22], and δ M ∆G* as a function of x2 is fitted to eq. [5.5.23] by nonlinear regression, with gA, K1, and K2 being treated as adjustable parameters.9 Figures 5.5.1 and 5.5.2 show some results.1 Clearly eq. [5.5.24] possesses the functional flexibility to describe the data. (In some systems a 1-step (2-state) equation is adequate. To transform eq. [5.5.24] to a 1-step version, set K2 = 0 and let γ ′ = γ 2 − γ 1.) The next step is to examine the parameter values for their possible physical significance. It seems plausible that K1 and K2 should be larger than unity, but not “very large,” on the basis that the solutes are organic and so are the cosolvents, but the cosolvents are water-miscible so they are in some degree “water-like.” In fact, we find that nearly all K1 and K2 values fall between 1 and 15. Likewise the gA values seem, in the main, to be physically reasonable. Earlier estimates of g (reviewed in ref.1) put it in the range of 0.35-0.5. A itself can be estimated as the solvent-accessible surface area of the solute, and many of the gA values found were consistent with such estimates, though some were considerably smaller than expected. Since gA arises in the theory as a hydrophobicity parameter, it seemed possible that A in the equation represents only the nonpolar surface area of the


287

molecule. An experiment whose results are summarized in Table 5.5.1 was designed to examine this possibility.10 Table 5.5.1. Surface area estimates of biphenyls10 Compound

Atotal

Anonpolar

gA

Biphenyl

179 (3)

179 (3)

74 (0.6)

4-Hydroxybiphenyl

185 (4)

155 (7)

69 (0.3)

4,4’-Dihydroxybiphenyl

203 (5)

126 (7)

53 (0.5)

4-Bromobiphenyl

217 (7)

217 (7)

87 (1.4)

Areas in 2 molecule; standard deviations in parentheses. The cosolvent was methanol.

Evidently the experimental gA estimate is correlated with Anonpolar rather than with Atotal, and the linear relationship yields the estimate g = 0.37. On the other hand, there was evidence1 that g depends upon cosolvent identity (for a given solute), and LePree and Mulski8,11 examined this possibility. Their findings led to an empirical but quite general correlation between gA and properties of the cosolvent and solute: gA = −42 log PM + 11log PR

[5.5.25]

In eq. [5.5.25] PM is the 1-octanol/water partition coefficient of the pure organic cosolvent and PR is the partition coefficient of the solute. Table 5.5.2 gives examples of the application of eq. [5.5.25]. Table 5.5.2 Experimental and calculated gA values8,11

Solute

Solvent

gA, 2 molecule-1 Calculated

Observed

Naphthalene

Methanol

66

63

Naphthalene

Ethanol

48

54

Naphthalene

2-Propanol

35

43

Naphthalene

1,2-Propanediol

92

71

Naphthalene

1,2-Ethanediol

116

102

Naphthalene

Acetone

45

69

Naphthalene

DMSO

120

127

4-Nitroaniline

Methanol

46

35

4-Nitroaniline

Ethanol

29

21

4-Nitroaniline

2-Propanol

15

11

4-Nitroaniline

1,2-Ethanediol

96

84

4-Nitroaniline

Acetone

25

37

288

Kenneth A. Connors

Solute

gA, 2 molecule-1

Solvent

Calculated

Observed

4-Nitroaniline

DMSO

101

86

4-Nitroaniline

Acetonitrile

30

29

We now encounter a curious observation. The parameter gA is constrained, in the nonlinear regression fitting program, to be constant as x2 varies over its entire range from 0 to 1. We have identified A as the nonpolar surface area of the solute (though it may actually be the corresponding area of the solvent cavity, and so may show some cosolvent dependency, which we have ignored). The quantity g can then be estimated. For example, from Table 5.5.2 for naphthalene (A = 147 2 molecule-1), g varies from 0.29 (for 2-propanol) to 0.86 (for DMSO). Yet how can g possess these different values in different cosolvents, maintain its constancy as x2 varies, and then collapse to the unique value it must possess in water, the reference solvent for all systems? An independent calculation gives g = 0.41±0.03 in water.8 Some tentative explanations for this puzzle have been offered,8 and we return to this issue in Section 5.5.4. Turning to the K1 and K2 parameters, we have observed that these are relatively insensitive to the identity of the solute, but that they depend upon the cosolvent, whose polarity is a controlling factor. Table 5.5.3 gives some empirical correlations that provide routes to the prediction of K1 and K1K2. In Table 5.5.3, PM is the 1-octanol/water partition coefficient of the pure cosolvent,12 and ET is the Dimroth-Reichardt solvatochromic polarity parameter.13 We thus have the capability of predicting gA, K1, and K1K2, which extends the utility of eq. [5.5.23] from the merely descriptive to the predictive. Table 5.5.3. Empirical relationships for estimating solvation constants Equation

n

r

Restrictions

log K1 = -0.0316 ET + 2.24

10

0.91

-

log (K1/K2) = 0.0171 ET - 9.23

4

0.98

ET > 51

log (K1/K2) = -0.0959 ET + 4.60

6

0.85

ET < 52

log K1K2 = 1.31 log PM + 1.81

6

0.91

log PM > -1.0

5.5.3.2 Surface tension In the development of the basic phenomenological model, eq. [5.5.23], we derived a relationship for the surface tension of the solvation shell. Combining eqs. [5.5.16] and [5.5.18] yields  K x x + 2K1K 2 x 22  γ = γ1 + γ′  2 1 1 2 2   x 1 + K1 x 1 x 2 + K1K 2 x 2 

[5.5.26]

where γ ′ = ( γ 2 − γ 1 ) / 2. Now if we identify the solute-solvation shell system with the air-solvent interface, we are led to test eq. [5.5.26] as a description of the composition dependence


Figure 5.5.3. Surface tension of 2-propanol-water mixtures. The smooth line is drawn with eq. 5.5.26. (Reproduced with permission from the Journal of Solution Chemistry. reference 14.)

289

Figure 5.5.4. Surface tension of glycerol-water mixtures. The smooth line is drawn with eq. 5.5.26. (Reproduced with permission from the Journal of Solution Chemistry. reference 14.)

of surface tension in mixed solvent systems, air playing the role of the solute. Figures 5.5.3 and 5.5.4 show examples of these curve-fits.14 When K2 = 0, eq [5.5.26] gives the 1-step model, eq. [5.5.27], where γ ′ = γ 2 − γ 1.  K1 x 2  γ = γ1 + γ′    x 1 + K1 x 2 

[5.5.27]

We had earlier15 published an equation describing the dependence of surface tension on composition, and a comparison of the two approaches has been given;14 here we will restrict attention to eqs. [5.5.26] and [5.5.27]. Suppose we set K1 = 2 and K2 = 1/2 in eq. [5.5.26]. This special condition converts eq. [5.5.26] to γ = γ1 x1 + γ 2 x 2

[5.5.28]

which corresponds to ideal behavior; the surface tension is a linear function of x2. The restriction K1 = 2, K2 = 1/2 is, however, a unique member of a less limited special case in which K1 = 4K2. This important condition (except when it happens to occur fortuitously) implies the existence of two identical and independent binding sites.16 Inserting K2 = K1/4 into eq. [5.5.26] yields, upon simplification, eq. [5.5.29], where γ ′ = γ 2 − γ 1.

290

Kenneth A. Connors

 (K1 / 2)x 2  γ = γ1 + γ′    x 1 + (K1 / 2)x 2 

[5.5.29]

Eq. [5.5.29] will be recognized as equivalent to eq. [5.5.27] for the 1-step model. The interpretation is as follows: the 2-step model, eq. [5.5.26], can always be applied, but if the result is that K1 ≈ 4K2 the 1-step model will suffice to describe the data. Moreover, if K1 ≈ 4K2 from the 2-step treatment, no physical significance is to be assigned to the second parameter. These considerations are pertinent to real systems. Table 5.5.4 lists K1 and K2 values obtained by applying eq. [5.5.26] to literature data.15 Several systems conform reasonably to the K1 ≈ 4K2 condition. Recall that ideal behavior requires the special case K1 = 2, K2 = 1/2. The less restrictive condition K1 ≈ 4K2 we call “well-behaved.” Figure 5.5.3 shows a well-behaved system; Figure 5.5.4 shows one that is not well-behaved. The distinction is between a hyperbolic dependence on x2 (well-behaved) and a non-hyperbolic dependence. Table 5.5.4. Solvation parameter estimates for surface tension data according to eq. [5.5.26] K1

K2

K1/K2

Methanol

19.8

2.9

6.8

2-Propanol

130

29.4

4.4

1-Propanol

232

50

4.6

t-Butanol

233

65

3.6

Acetic acid

115

2.7

42.6

Acetone

138

7.1

19.4

Acetonitrile

33.4

14.5

2.3

Dioxane

62.1

7.4

8.4

THF

136

25.9

5.3

Glycerol

22.7

0.80

28.4

DMSO

12.3

1.43

8.6

Formamide

5.52

2.57

2.1

Ethylene glycol

9.4

2.62

3.6

Cosolvent

A further observation from these results is that some of the K1 values are much larger than those encountered in solubility studies. Correlations with log PM have been shown.14 5.5.3.3 Electronic absorption spectra The energy of an electronic transition is calculated from the familiar equation E T = hv =

hc λ

[5.5.30]


291

where h is Planck’s constant, c is the velocity of light, v is frequency, and λis wavelength. If λ is expressed in nm, eq. [5.5.31] yields ET in kcal mol-1. E T = 2.859 × 10 4 / λ

[5.5.31]

The phenomenological theory has been applied by Skwierczynski to the ET values of the Dimroth-Reichardt betaine,13 a quantity sensitive to the polarity of the medium.17 The approach is analogous to the earlier development. We need only consider the solvation effect. The solute is already in solution at extremely low concentration, so solute-solute interactions need not be accounted for. The solvent cavity does not alter its size or shape during an electronic transition (the Franck-Condon principle), so the general medium effect does not come into play. We write ET of the mixed solvent as a weighted average of contributions from the three states: E T (x 2 ) = FWW E T (WW ) + FWM E T (WM) + FMM E T (MM)

[5.5.32]

where the symbolism is obvious. Although ET(WW) can be measured in pure water and ET(MM) in pure cosolvent, we do not know ET(WM), so provisionally we postulate that ET (WM) = [ET(WW) + ET(MM)]/2. Defining a quantity Γ by Γ=

E T (x 2 ) − E T (WW )

E T (MM) − E T (WW )

[5.5.33]

we find, by combining eqs. [5.5.9], [5.5.10], and [5.5.32], Γ=

K1 x 1 x 1 / 2 + K1K 2 x 22 x 12 + K1 x 1 x 2 + K1K 2 x 22

[5.5.34]

The procedure is to fit Γ to x2. As before, a 1-parameter version can be obtained by setting K2 = 0: Γ=

K1 x 2 x 1 + K1 x 2

[5.5.35]

Figure 5.5.5 shows a system that can be satisfactorily described by eq. [5.5.35], whereas the system in Figure 5.5.6 requires eq. [5.5.34]. The K1 values are similar in magnitude to those observed from solubility systems, with a few larger values; K2, for those systems requiring eq. [5.5.34], is always smaller than unity. Some correlations were obtained of K1 and K2 values with solvent properties. Figure 5.5.7 shows log K1 as a function of log PM, where PM is the partition coefficient of the pure organic solvent. 5.5.3.4 Complex formation. We now inquire into the nature of solvent effects on chemical equilibria, taking noncovalent molecular complex formation as an example. Suppose species S (substrate) and L (ligand) interact in solution to form complex C, K11 being the complex binding constant. S

+ L

K11

C

[5.5.36]

292

Kenneth A. Connors

Figure 5.5.5. Dependence of ET on composition for the methanol-water system. The smooth line was drawn with eq. 5.5.35. (Reproduced with permission from the Journal of the Chemical Society. Perkin Transactions 2, reference 17.)

Figure 5.5.7. A plot of log K1 from the ET data against log PM; the circles represent 1-step solvents (eq. 5.5.35) and the squares, 2-step solvents (eq. 5.5.34). (Reproduced with permission from the Journal of the Chemical Society, Perkin Transactions 2, reference 17.)

Figure 5.5.6. Dependence of ET on composition for the acetone-water system. The smooth line was drawn with eq. 5.5.34. (Reproduced with permission from the Journal of the Chemical Society. Perkin Transactions 2,. reference 17.)

It is at once evident that this constitutes a more complicated problem than those we have already considered inasmuch as here we have three solutes. We begin with the thermodynamic cycles shown as Figure 5.5.8; these cycles describe complex formation in the solid, solution, and gas phases horizontally, and the energy changes associated with the indicated processes. ∆Glatt corresponds to the crystal lattice energy (solute-solute interactions), ∆Gcav represents the energy of cavity formation (identical with the general medium effect of Section 5.5.2). ∆Gcomp is the free energy of complex formation, which in the solution phase is given by eq. [5.5.37]. ∆G comp (l ) = −kT ln K11 [5.5.37]

Eq. [5.5.37] gives the free energy with respect to a 1M standard state, because the unit of K11 is M-1. To calculate the unitary (mole fraction) free energy change we write, instead of eq. [5.5.37], eq. [5.5.38]:

(

∆G comp (l ) = −kT ln K11M *ρ

)

[5.5.38]

where M* is the number of moles of solvent per kg of solvent and ρ is the solution density. The unitary free energy does not include the entropy of mixing. From cycle gl in Figure 5.5.8 we obtain eq. [5.5.39].


293

Figure 5.5.8. Thermodynamic cycles for bimolecular association. The symbols s, l, g represent solid, liquid, and gas phases; the superscripts refer to substrate S, ligand L, and complex C. (Reproduced with permission from the Journal of Solution Chemistry, reference 18.)

(

)

C C ∆G comp (g) + ∆G cav + ∆Gsolv − ∆G comp (l )

(

) (

)

L L S − ∆G cav + ∆Gsolv − ∆G Scav + ∆Gsolv =0

[5.5.39]

We apply the δM operator to eq. [5.5.39] L C S + δM ∆Gsolv − δM ∆Gsolv δ M ∆G Ccav − δ M ∆G Scav − δM ∆G cav L − δ M ∆Gsolv = δ M ∆G comp (l )

[5.5.40]

where we have assumed δM∆Gcomp (g) = 0, which is equivalent to supposing that the structure of the complex (the spatial relationship of S and L) does not depend upon solvent composition, or that the intersolute effect is composition independent. Also applying the δM treatment to eq. [5.5.4] gives δ M ∆Gsoln = δ M ∆G cav + δM ∆Gsolv

[5.5.41]

for each species; recall that ∆Ggen med and ∆Gcav are identical. Use eq. [5.5.41] in [5.5.40]: C L S δ M ∆G comp (l ) = δ M ∆Gsoln − δM ∆Gsoln − δM ∆Gsoln

[5.5.42]

Eq. [5.5.42] says that the solvent effect on complex formation is a function solely of the solvent effects on the solubilities of reactants (negative signs) and product (positive sign). This is a powerful result, because we already have a detailed expression, eq. [5.5.23], for each of

294

Kenneth A. Connors

the three quantities on the right-hand side of eq. [5.5.42]. Thus the problem is solved in principle.18 In practice, of course, there are difficulties. Each of the δM∆Gsoln terms contains three adjustable parameters, for nine in all, far too many for eq. [5.5.42] to be practicable in that form. We therefore introduce simplifications in terms of some special cases. The first thing to do is to adopt a 1-step model by setting K2 = 0. This leaves a six-parameter equation, which, though an approximation, will often be acceptable, especially when the experimental study does not cover a wide range in solvent composition (as is usually the case). This simplification gives eq. [5.5.43]. δ M ∆G *comp = −

( gAC γ ′ − kT ln K1C )K1C x 2 x 1 + K1C x 2

( gA L γ ′ − kT ln K1L )K1L x 2 x 1 + K1L x 2

−

( gAS γ ′ − kT ln K1S )K1S x 2 x 1 + K1S x 2 [5.5.43]

Next, in what is labeled the full cancellation approximation, we assume K1C = K1S = K1L = K1 and we write ∆gA = gAC - gAS - gAL. The result is δ M ∆G *comp =

(kT ln K1 + ∆gAγ′)K1 x 2 x 1 + K1 x 2

[5.5.44]

and we now have a 2-parameter model. The assumption of identical solvation constants is actually quite reasonable; recall from the solubility studies that K1 is not markedly sensitive to the solute identity. The particular example of cyclodextrin complexes led to the identification of another special case as the partial cancellation approximation; in this case we assume K1C = K1S < K1L, and the result is, approximately.19 δ M ∆G *comp =

(kT ln K1 − gAγ′)K1 x 2 x 1 + K1 x 2

[5.5.45]

Functionally eqs. [5.5.44] and [5.5.45] are identical; the distinction is made on the basis of the magnitudes of the parameters found. Note that gA in eq. [5.5.45] is a positive quantity whereas ∆gA in eq. [5.5.44] is a negative quantity. In eq. [5.5.45] it is understood that gA and K1 refer to L. Eqs. [5.5.44] and [5.5.45] both have the form δ M ∆G *comp =

(kT ln K1 + Gγ′)K1 x 2 x 1 + K1 x 2

[5.5.46]

where G = ∆gA in eq. [5.5.44] and G = -gA in eq. [5.5.45]. Table 5.5.5 shows G and K1 values obtained in studies of α-cyclodextrin complexes.19,20 The assignments are made on the basis of the magnitude of K1; those values substantially higher than typical solubility K1 values suggest that the full cancellation condition is not satisfied. After the assignments are made, G can be interpreted as either ∆gA (full cancellation) or -gA (partial cancellation).


295

Notice in Table 5.5.5 that all full cancellation systems give substantial negative ∆gA values. If g is constant, ∆gA = g∆A, and the negative ∆A value leads to a solvophobic driving force of g∆Aγ for complex formation. (The dioxane system in Table 5.5.5 is unassigned because its K1 value suggests partial cancellation whereas its G value suggests full cancellation). Table 5.5.5. Parameter values of the 4-nitroaniline/α-cyclodextrin and methyl orange/ α-cyclodextrin systems19,20 Cosolvent

K1

Ga

Cancellation assignment

4-Nitroaniline Acetonitrile

55

+3

Partial

2-Propanol

46

-3

Partial

Ethanol

29

-9

Partial

Acetone

10

-57

Full

Methanol

3.1

-68

Full

Methyl orange Acetone

46

-3

Partial

2-Propanol

43

-11

Partial

Acetonitrile

40

-13

Partial

Dioxane

31

-38

(Unassigned)

Ethylene glycol

7.7

-58

Full

DMSO

6.4

-66

Full

Methanol

4.9

-43

Full

Units are 2 molecule-1

a

5.5.3.5 Chemical kinetics Treatment of the solvent effect on chemical reaction rates by means of the phenomenological theory is greatly facilitated by the transition state theory, which postulates that the initial and transition states are in (virtual) equilibrium. Thus the approach developed for complex formation is applicable also to chemical kinetics. Again we begin with a thermodynamic cycle, Figure 5.5.9, where R represents the reactant (initial state) in a unimolecular reaction, R is the transition state, and P is the product. From Figure 5.5.9 we write eq. Figure 5.5.9. Thermodynamic cycle for a unimolecular [5.5.47], where ∆Grxn(1) subsequently writreaction. (Reproduced with permission from the Journal of Pharmaceutical Sciences, reference 21.)

296

Kenneth A. Connors

ten ∆Grxn, is the free energy of activation in the solution phase. R R ∆Grxn(1) = ∆Grxn(g) + (∆Ggen med + ∆Gsolv) − (∆G gen med + ∆Gsolv )

[5.5.47]

Applying the δM operation gives eq. [5.5.48]: δ M ∆Grxn = ∆Grxn(x2) − ∆Grxn(x2=0)

[5.5.48]

The quantity ∆Grxn(g), disappears in this subtraction, as do other composition-independent quantities. We make use of eq. [5.5.13] and eq. [5.5.19] to obtain a function having six parameters, namely K1R, K2R, K1, gAR, and gA. This function is made manageable by adopting the full cancellation approximation, setting K1R = K1 = K1 and K2R = K2 = K2. We then obtain

δ M ∆G

rxn

=

(∆gA γ ′K1x1x2 + 2∆gA γ ′K1K2x 22 )/(x12 + K1x1x2 + K1K2x 22 )

[5.5.49]

where ∆gA = gA - gAR; this is the difference between the curvature-corrected molecular surface areas of the cavities containing the transition state and the reactant. This quantity may be positive or negative. LePree21 tested eq. [5.5.49] with the decarboxylative dechlorination of N-chloroamino acids in mixed solvents RCH(NHCl)COOH + H2O → RCHO + NH3 + HCl + CO2

Figure 5.5.10. Solvent effect on the decomposition of N-chloroaniline in acetonitrile-water mixtures. The smooth curve is drawn with eq. 5.5.49. (Reproduced with permission from the Journal of Pharmaceutical Sciences, reference 21.)

[5.5.50]

Figure 5.5.11. Solvent effect on the decomposition of N-chloroleucine in 2-propanol-water mixtures. The smooth curve is drawn with eq. 5.5.49. (Reproduced with permission from the Journal of Pharmaceutical Sciences, reference 21.)


297

For this test, the reaction possesses these very desirable features: (1) the kinetics are first order, and the rate-determining step is unimolecular; (2) the reaction rate is independent of pH over the approximate range 4-13; (3) the rate-determining step of the process is not a solvolysis, so the concentration of water does not appear in the rate equation; and (4) the reaction is known to display a sensitivity to solvent composition. Figures [5.5.10] and [5.5.11] show curve-fits, and Tables 5.5.6 and 5.5.7 give the parameter values obtained in the curve-fitting regression analysis. Observe that ∆gA is positive. This means that the transition state occupies a larger volume than does the reactant. This conclusion has been independently confirmed by studying the pressure dependence of the kinetics.21,22 Table 5.5.6. Model parameters for solvent effects on the decomposition of N-chloroalanine21 K1

K2

∆gA, 2 molecule-1

Methanol

2.8

3.1

16.8

Ethanol

5.7

2.8

23.8

1-Propanol

13.0

3.2

22.0

2-Propanol

5.9

11

20.6

Ethylene glycol

4.4

1.7

34.8

Propylene glycol

5.2

6.0

27.0

Acetonitrile

7.0

2.0

42

Dioxane

8.9

4.3

55

Cosolvent

Table 5.5.7. Model parameters for solvent effects on the decomposition of N-chloroleucine21 K1

K2

∆gA, 2 molecule-1

Methanol

2.5

4.0

19.3

2-Propanol

2.7

40

24.2

Ethylene glycol

4.4

3.2

38

Acetonitrile

7.5

3.9

43

Cosolvent

The success shown by this kinetic study of a unimolecular reaction unaccompanied by complications arising from solvent effects on pH or water concentration (as a reactant) means that one can be confident in applying the theory to more complicated systems. Of course, an analysis must be carried out for such systems, deriving the appropriate functions and making chemically reasonable approximations. One of the goals is to achieve a practical level of predictive ability, as for example we have reached in dealing with solvent effects on solubility.

298

Kenneth A. Connors

5.5.3.6 Liquid chromatography In reverse phase high-pressure liquid chromatography (RP HPLC), the mobile phase is usually an aqueous-organic mixture, permitting the phenomenological theory to be applied. LePree and Cancino23 carried out this analysis. The composition-dependent variable is the capacity factor k′, defined by eq. [5.5.51], k′ =

VR − VM t R − t M = VM tM

[5.5.51]

where VR is the retention volume of a solute, tR is the retention time, VM is the column dead volume (void volume), and tM is the dead time. It is seldom possible in these systems to use pure water as the mobile phase, so LePree reversed the usual calculational procedure, in which water is the reference solvent, by making the pure organic cosolvent the reference. This has the effect of converting the solvation constants K1 and K2 to their reciprocals, but the form of the equations is unchanged. For some solvent systems a 1-step model was adequate, but others required the 2-step model. Solvation constant values (remember that these are the reciprocals of the earlier parameters with these labels) were mostly in the range 0.1 to 0.9, and the gA values were found to be directly proportional to the nonpolar surface areas of the solutes. This approach appears to offer advantages over earlier theories in this application because of its physical significance and its potential for predicting retention behavior. 5.5.4 INTERPRETATIONS The very general success of the phenomenological theory in quantitatively describing the composition dependence of many chemical and physical processes arises from the treatment of solvation effects by a stoichiometric equilibrium model. It is this model that provides the functional form of the theory, which also includes a general medium effect (interpreted as the solvophobic effect) that is functionally coupled to the solvation effect. The parameters of the theory appear to have physical significance, and on the basis of much experimental work they can be successfully generated or predicted by means of empirical correlations. The theory does not include molecular parameters (such as dipole moments or polarizabilities), and this circumstance deprives it of any fundamental status, yet at the same time enhances its applicability to the solution of practical laboratory problems. Notwithstanding the widespread quantitative success of the theory, however, some of the observed parameter values have elicited concern about their physical meaning, and it is to address these issues, one of which is mentioned in 5.5.3.1, that the present section is included. 5.5.4.1 Ambiguities and anomalies Consider a study in which the solubility of a given solute (naphthalene is the example to be given later) is measured in numerous binary aqueous-organic mixed solvent systems, and eq. [5.5.23] is applied to each of the mixed solvent systems, the solvent effect δ M ∆G*soln being calculated relative to water (component 1) in each case. According to hypothesis, the parameter gA is independent of solvent composition. This presumably means that it has the same value in pure solvent component 1 and in pure solvent component 2, since it is supposed not to change as x2 goes from 0 to 1. And in fact the nonlinear regression analyses support the conclusion that gA is a parameter of the system, independent of composition.


299

But in the preceding paragraph no restriction has been placed on the identity of solvent component 2, so the conclusion must apply to any cosolvent 2 combined with the common solvent 1, which is water. This means that all mixed solvent systems in this study as described should yield the same value of gA. But this is not observed. Indeed, the variation in gA can be extreme, in a chemical sense; see Table 5.5.2. This constitutes a logical difficulty. There is another anomaly to be considered. In nearly all of the nonelectrolyte solubility data that have been subjected to analysis according to eq. [5.5.23] the solute solubility increases as x2, the organic cosolvent concentration, increases, and gA is positive, the physically reasonable result. But in the sucrose-water-ethanol system, the sucrose solubility decreases as x2 increases, and gA is negative. There appears to be no physically reasonable picture of a negative gA value. A further discrepancy was noted in 5.5.3.2, where we saw that some of the solvation constants evaluated from surface tension data did not agree closely with the corresponding numbers found in solubility studies. 5.5.4.2 A modified derivation Recognizing that the original condition that g and A are independent of composition was unnecessarily restrictive, we replace eq. [5.5.18] with eq. [5.5.52], where the subscripts 1 and 2 indicate values in the pure solvents 1 and 2. áγ = g1 A1 γ 1 + (g 2 A2 γ 2 − g1 A1 γ 1 )f 2

[5.5.52]

It is important for the moment to maintain a distinction between gA in the original formulation, a composition-independent quantity, and áÂ in eq. [5.5.52], a composition-dependent quantity. Eq. [5.5.52] combines the composition dependence of three entities into a single grouping, áÂγ, which is probably an oversimplification, but it at least generates the correct values at the limits of x2 = 0 and x2 = 1; and it avoids the unmanageable algebraic complexity that would result from a detailed specification of the composition dependence of the three entities separately. Eq. [5.5.14] now is written ∆Ggen med = áÂγ and development as before yields eq. [5.5.53] as the counterpart to eq. [5.5.23], where δMgAγ = g2A2γ2 g1A1γ1. δM ∆G*soln =

(δM gAγ / 2 − kT ln K1)K1x1x 2 + (δM gAγ − kT ln K1K2 )K1K2 x 22 x12 + K1x1x 2 + K1K2 x 22

[5.5.53]

Comparison of eqs. [5.5.23] and [5.5.53] gives eq. [5.5.54], which constitutes a specification of the meaning of gA in the original formulation in terms of the modified theory. gA(γ 2 − γ 1 ) = g 2 γ 2 A2 − g1 γ 1 A1

[5.5.54]

Now, the right-hand side of eq. [5.5.54] is a constant for given solute and solvent system, so the left-hand side is a constant. This shows why gA in the original theory (eq. [5.5.23]), is a composition-independent parameter of the system. Of course, in the derivation of eq. [5.5.23] gA had been assumed constant, and in effect this assumption led to any composition dependence of gA being absorbed into γ. In the modified formulation we acknowledge

300

Kenneth A. Connors

that the composition dependence of the product áÂγ is being accounted for without claiming that we can independently assign composition dependencies to the separate factors in the product. 5.5.4.3 Interpretation of parameter estimates Eq. [5.5.54] constitutes the basis for the resolution of the logical problem, described earlier, in which different cosolvents, with a given solute, yield different gA values, although gA had been assumed to be independent of composition. As eq. [5.5.54] shows, gA is determined by a difference of two fixed quantities, thus guaranteeing its composition independence, and at the same time permitting gA to vary with cosolvent identity. Eq. [5.5.53] is therefore conceptually sounder and physically more detailed than is eq. [5.5.23]. Eq. [5.5.53] shows, however, that in the absence of independent additional information (that is, information beyond that available from the solubility study alone) it is not possible to dissect the quantity (g2γ2A2 - g1γ1A1) into its separate terms. In some cases such additional information may be available, and here we discuss the example of naphthalene solubility in mixed aqueous-organic binary mixtures. Table 5.5.8 lists the values of gA(γ2-γ1) obtained by applying eq. [5.5.23] to solubility data in numerous mixed solvent systems.8 In an independent calculation, the solubility of naphthalene in water was written as eq. [5.5.55], ∆G*soln (x 2 = 0) = ∆G cryst + g1 A1 γ 1

[5.5.55]

which is equivalent to eq. [5.5.4]. ∆Gcryst was estimated by conventional thermodynamic arguments and ∆Gsolv was omitted as negligible,24 yielding the estimate g1A1γ1 = 4.64 x 10-20 J molecule-1. With eq. [5.5.54] estimates of g2A2γ2 could then be calculated, and these are listed in Table 5.5.8. Table 5.5.8. Parameter estimates and derived quantities for naphthalene solubility in water-cosolvent mixtures at 25oCa γ2, erg cm-2

1020 gA(γ2-γ1), J molecule-1

1020 g2A2γ2, J molecule-1

Methanol

22.4

-3.11

+1.53

Ethanol

21.8

-2.70

1.94

Isopropanol

20.8

-2.19

2.45

Propylene glycol

37.1

-2.46

2.18

Ethylene glycol

48.1

-2.24

2.22

Acetone

22.9

-3.37

1.27

Dimethylsulfoxide

42.9

-3.67

0.97

Cosolvent

Data from ref. (8); γ 1 = 71.8 erg cm-2, g1A1γ 1 = 4.64 x 10-20 J molecule-1.

a

Observe that g1A1γ1 and g2A2γ2 are positive quantities, as expected; gA(γ2-γ1) is negative because of the surface tension difference. It is tempting to divide each of these quantities by its surface tension factor in order to obtain estimates of gA, g1A1, and g2A2, but this procedure may be unsound, as proposed subsequently.


301

5.5.4.4 Confounding effects Solute-solute interactions It is very commonly observed, in these mixed solvent systems, that the equilibrium solubility rises well above the dilute solution condition over some portion of the x2 range. Thus solution phase solute-solute interactions must make a contribution to ∆G*soln . To some extent these may be eliminated in the subtraction according to eq. [5.5.22], but this operation cannot be relied upon to overcome this problem. Parameter estimates may therefore be contaminated by this effect. On the other hand, Khossravi25 has analyzed solubility data for biphenyl in methanol-water mixtures by applying eq. [5.5.23] over varying ranges of x2; he found that gA(γ2-γ1) was not markedly sensitive to the maximum value of x2 chosen to define the data set. In this system the solubility varies widely, from x3 = 7.1 x 10-7 (3.9 x 10-5 M) at x2 = 0 to x3 = 0.018 (0.43 M) at x2 = 1. Coupling of general medium and solvation effects In this theory the general medium and solvation effects are coupled through the solvation exchange constants K1 and K2, which determine the composition of the solvation shell surrounding the solute, and thereby influence the surface tension in the solvation shell. But the situation is actually more complicated than this, for if surface tension-composition data are fitted to eq. [5.5.26] the resulting equilibrium constants are not numerically the same as the solvation constants K1 and K2 evaluated from a solubility study in the same mixed solvent. Labeling the surface tension-derived constants K1′ and K′2 , it is usually found that K1′ >K1 and K′2 >K2. The result is that a number attached to γ at some x2 value as a consequence of a nonlinear regression analysis according to eq. [5.5.23] will be determined by K1 and K2, and this number will be different from the actual value of surface tension, which is described by K′1 and K′2 . But of course the actual value of γ is driving the general medium effect, so the discrepancy will be absorbed into gA. The actual surface tension (controlled by K1′ and K′2 ) is smaller (except when x2 = 0 and x2 = 1) than that calculated with K1 and K2. Thus gapparent = gtrue x γ(K′1,K′2 )/γ(K1,K2). This effect will be superimposed on the curvature correction factor that g represents, as well as the direct coupling effect of solvation mentioned above. The cavity surface area In solubility studies of some substituted biphenyls, it was found (see 5.5.3.1) that gA evaluated via eq. [5.5.23] was linearly correlated with the nonpolar surface area of the solutes rather than with their total surface area; the correlation equation was gA = 0.37 Anonpolar. It was concluded that the A in the parameter gA is the nonpolar surface area of the solute. This conclusion, however, was based on the assumption that g is fixed. But the correlation equation can also be written gA = 0.37 FnonpolarAtotal, where Fnonpolar = Anonpolar/Atotal is the fraction of solute surface area that is nonpolar. Suppose it is admitted that g may depend upon the solute (more particularly, it may depend upon the solute’s polarity); then the correlation is consistent with the identities A = Atotal and g = 0.37 Fnonpolar. Thus differences in gA may arise from differences in solute polarity, acting through g. But A may itself change, rather obviously as a result of solute size, but also as a consequence of change of solvent, for the solvent size and geometry will affect the shape and size of the cavity that houses the solute.

302

Kenneth A. Connors

The role of interfacial tension In all the preceding discussion of terms having the gAγ form, γ has been interpreted as a surface tension, the factor g serving to correct for the molecular-scale curvature effect. But a surface tension is measured at the macroscopic air-liquid interface, and in the solution case we are actually interested in the tension at a molecular scale solute-solvent interface. This may be more closely related to an interfacial tension than to a surface tension. As a consequence, if we attempt to find (say) g2A2 by dividing g2A2γ2 by γ2, we may be dividing by the wrong number. To estimate numbers approximating to interfacial tensions between a dissolved solute molecule and a solvent is conjectural, but some general observations may be helpful. Let γX and γY be surface tensions (vs. air) of pure solvents X and Y, and γXY the interfacial tension at the X-Y interface. Then in general, γ XY = γ X + γ Y − W XY − WYX

[5.5.56]

where WXY is the energy of interaction (per unit area) of X acting on Y and WYX is the energy of Y acting on X. When dispersion forces alone are contributing to the interactions, this equation becomes26

(

γ XY = γ X + γ Y − 2 γ Xd γ Yd

)

1/ 2

[5.5.57]

where γXd and γYd are the dispersion force components of γX and γY. In consequence, γXY is always smaller than the larger of the two surface tensions, and it may be smaller than either of them. Referring now to Table 5.5.8, if we innocently convert g2A2γ2 values to estimates of g2A2 by dividing by γ2, we find a range in g2A2 from 23 Å2 molecule-1 (for dimethylsulfoxide) to 118 Å2 molecule-1 (for isopropanol). But if the preceding argument is correct, in dividing by γ2 we were dividing by the wrong value. Taking benzene (γ = 28 erg cm-2) as a model of supercooled liquid naphthalene, we might anticipate that those cosolvents in Table 5.5.8 whose γ2 values are greater than this number will have interfacial tensions smaller than γ2, hence should yield g2A2 estimates larger than those calculated with γ2, and vice versa. Thus, the considerable variability observed in g2A2 will be reduced. On the basis of the preceding arguments it is recommended that gAγ terms (exemplified by g1A1γ1, g2A2γ2, and gA(γ2-γ1)) should not be factored into gA quantities through division by γ, the surface tension, (except perhaps to confirm that magnitudes are roughly as expected). This conclusion arises directly from the interfacial tension considerations. Finally let us consider the possibility of negative gA values in eq. [5.5.23]. Eq. [5.5.54] shows that a negative gA is indeed a formal possibility, but how can it arise in practice? We take the water-ethanol-sucrose system as an example; gA was reported to be negative for this system. Water is solvent 1 and ethanol is solvent 2. This system is unusual because of the very high polarity of the solute. At the molecular level, the solute in contact with these solvents is reasonably regarded as supercooled liquid sucrose, whose surface tension is unknown, but might be modeled by that of glycerol (γ = 63.4 erg cm-2). In these very polar systems capable of hydrogen-bonding eq. [5.5.57] is not applicable, but we can anticipate that the sucrose-water interaction energies (the WXY and WXY terms in eq. [5.5.56] are


303

greater than sucrose-ethanol energies. We may expect that the sucrose-water interfacial tension is very low. Now, gA turned out to be negative because gA(γ2-γ1), a positive quantity as generated by eq. [5.5.23], was divided by (γ2-γ1), a difference of surface tensions that is negative. Inevitably gA was found to be negative. The interfacial tension argument, however, leads to the conclusion that division should have been by the difference in interfacial tensions. We have seen that the interfacial tension between sucrose and water may be unusually low. Thus the factor (γ2-γ1), when replaced by a difference of interfacial tensions, namely [γ(sucrose/ethanol) - γ(sucrose/water)], is of uncertain magnitude and sign. We therefore do not know the sign of gA; we only know that the quantity we label gA(γ2-γ1) is positive. This real example demonstrates the soundness of the advice that products of the form gAγ not be separated into their factors.27,28 5.5.5 NOTES AND REFERENCES 1 2 3 4 5 6 7 8 9

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

25 26 27

28

D. Khossravi and K.A. Connors, J. Pharm. Sci., 81, 371 (1992). R.R. Pfeiffer, K.S. Yang, and M.A. Tucker, J. Pharm. Sci., 59, 1809 (1970). J.B. Bogardus, J. Pharm. Sci., 72, 837 (1983). P.L. Gould, J.R. Howard, and G.A. Oldershaw, Int. J. Pharm., 51, 195 (1989). Also, when K1 = 1 and K2 = 1, eq. [5.5.13] shows that ∆Gsolv = ∆GWW; in this special case the solvation energy is composition-independent. H.H. Uhlig, J. Phys. Chem., 41, 1215 (1937). J.E. Leffler and E. Grunwald, Rates and Equilibria of Organic Reactions, J. Wiley & Sons, New York, 1963, p. 22. J.M. LePree, M.J. Mulski, and K.A. Connors, J. Chem. Soc., Perkin Trans. 2, 1491 (1994). The curvature correction factor g is dimensionless, as are the solvation constants K1 and K2. The parameter gA is expressed in Å2 molecule-1 by giving the surface tension the units J Å-2 (where 1 erg cm-2 = 1 x 10-23 J Å-2). D. Khossravi and K.A. Connors, J. Pharm. Sci., 82, 817 (1993). J.M. LePree, Ph.D. Dissertation, University of Wisconsin-Madison, 1995, p. 29. A. Leo, C. Hansch, and D. Elkins, Chem. Revs., 71, 525 (1971). C. Reichardt, Solvents and Solvent Effects in Organic Chemistry, VCH, Weinheim, 1988. D. Khossravi and K.A. Connors, J. Solution Chem., 22, 321 (1993). K.A. Connors and J.L. Wright, Anal. Chem., 61, 194 (1989). K.A. Connors, Binding Constants, Wiley-Interscience, New York, 1987, pp. 51, 78. R.D. Skwierczynski and K.A. Connors, J. Chem. Soc., Perkin Trans. 2, 467 (1994). K.A. Connors and D. Khossravi, J. Solution Chem., 22, 677 (1993). M.J. Mulski and K.A. Connors, Supramol, Chem., 4, 271 (1995). K.A. Connors, M.J. Mulski, and A. Paulson, J. Org. Chem., 57, 1794 (1992). J.M. LePree and K.A. Connors, J. Pharm. Sci., 85, 560 (1996). M.C. Brown, J.M. LePree, and K.A. Connors, Int. J. Chem. Kinetics, 28, 791 (1996). J.M. LePree and M.E. Cancino, J. Chromatogr. A, 829, 41 (1998). The validity of this approximation can be assessed. The free energy of hydration of benzene is given as -0.77 kJ mol-1 (E. Grunwald, Thermodynamics of Molecular Species, Wiley-Interscience, New York, 1997, p. 290). Doubling this to -1.5 kJ mol-1 because of the greater surface area of naphthalene and repeating the calculation gives g1A1γ 1 = 4.88 x 10-20 J molecule-1, not sufficiently different from the value given in the text to change any conclusions. D. Khossravi, Ph.D. Dissertation, University of Wisconsin-Madison, 1992, p. 141. F.M. Fowkes, Chemistry and Physics of Interfaces; American Chemical Society: Washington, D.C., 1965, Chap. 1. The introduction of the interfacial tension into the cavity term was first done by Yalkowsky et al.,28 who also argue that a separate solute-solvent interaction term is unneeded, as the solute-solvent interaction is already embodied in the interfacial tension. In our theory we explicitly show the coupling between the solute-solvent and solvent-solvent interactions (eq. [5.5.19]), but this is in addition to the solute-solvent interaction (eq. [5.5.13]). This difference between the two theories is a subtle issue that requires clarification. S.H. Yalkowsky, G.L. Amidon, G. Zografi, and G.L. Flynn, J. Pharm. Sci., 64, 48 (1975).

6

Swelling 6.1 MODERN VIEWS ON KINETICS OF SWELLING OF CROSSLINKED ELASTOMERS IN SOLVENTS E. Ya. Denisyuk Institute of Continuous Media Mechanics

V. V. Tereshatov Institute of Technical Chemistry Ural Branch of Russian Academy of Sciences, Perm, Russia

6.1.1 INTRODUCTION Diffusion phenomena encountered in mass-transfer of low-molecular liquids play an important role in many technological processes of polymer manufacture, processing, and use of polymeric materials. Diffusion of organic solvents in crosslinked elastomers may cause considerable material swelling. In this case, the polymeric matrix experiences strains as large as several hundred percent, while a non-homogeneous distribution of a liquid due to diffusion results in establishing stress-strain state capable of affecting the diffusion kinetics. The processes of material deformation and liquid diffusion in such systems are interrelated and nonlinear in nature and are strongly dependent on physical and geometrical nonlinearities. Therefore, exact relations of nonlinear mechanics of elastic-deformable continuum are the mainstream of a sequential theory of mass-transfer processes of low-molecular liquids in elastomers. The general principles of the development of nonlinear models of mass transfer in elastically deformed materials were developed in studies.1,2 The general formulation of constitutive equations and the use of non-traditional thermodynamic parameters such as partial stress tensors and diffusion forces lead to significant difficulties in attempts to apply the theory to the description of specific objects.3,4 Probably, because of this, the theory is little used for the solution of applied problems. In the paper,5 a theory for mechanical and diffusional processes in hyperelastic materials was formulated in terms of the global stress tensor and chemical potentials. The approach described in1,2 was used as the basic principle and was generalized to the case of a multi-component mixture. An important feature of the work5 is that, owing to the structure of constitutive equations, the general model can be used without difficulty to describe specific systems. In the paper6 the nonlinear theory5 was applied to steady swelling processes of crosslinked elastomers in solvents. The analytical and numerical treatment reveals three

306

E. Ya. Denisyuk, V. V. Tereshatov

possible mechano-diffusion modes which differ qualitatively. Self-similar solutions obtained for these modes describe asymptotic properties at the initial stage of swelling. These modes are related to thermodynamical material properties. The theoretical predictions have been verified in the experiments on real elastomers. 6.1.2 FORMULATION OF SWELLING FOR A PLANE ELASTOMER LAYER Consider an infinite plane elastomer layer of thickness 2h embedded in a low-molecular liquid. Suppose that the elastomer initially does not contain liquid and is unstrained. This state is taken as a reference configuration. Let us introduce the Cartesian coordinates (x,y,z) with the origin placed in the layer center and relate them to a polymer matrix. In the examined problem, the Cartesian coordinates will be used as the material coordinates. With reference to the layer, the x axis has a transverse direction and the other axes have longitudinal directions. In our approach, we define the problem under consideration as a one-dimensional problem, in which all quantities characterizing the elastomer state depend only on the x-coordinate. On swelling, the layer experiences transversal and longitudinal deformations which can be written as X = X ( x, t ) Y = ν(t )y

Z = ν(t )z

[6.1.1]

where (X,Y,Z) are the spatial Cartesian coordinates specifying the actual configuration of the polymeric matrix. From this it follows that the relative longitudinal stretch of the layer is λ 2 = λ 3 = ν(t) and the relative transversal stretch is λ1 = λ(x,t) = ∂X / ∂x. The quantity J = λ1 λ 2 λ 3 = λν 2

[6.1.2]

characterizes a local relative change in the material volume due to liquid absorption. The boundary conditions and the relations describing free swelling of the plane layer in the reference configuration are represented in5 as ∂N1 ∂  ∂N1  = D , N1 = N1 (x, t) ∂t ∂x  ∂x 

[6.1.3]

∂N2 / ∂t = 0

[6.1.4]

∂σ1 / ∂x = 0

[6.1.5]

N1 (x,0) = 0

[6.1.6]

∂N1 (0, t) / ∂x = 0, µ(h, t) = 0,

X (0, t) = 0

σ1 (h, t) = 0

σ 2 (x, t) = σ 3 (x, t) = 0

[6.1.7] [6.1.8] [6.1.9]

where: N1, N2 µ σk

the molar concentrations of the liquid and the chains of polymeric network of elastomer, respectively, the chemical potential of the liquid dissolved in material (k = 1,2,3) are the principal values of the Piola stress tensors.

6.1 Modern views on kinetics of swelling

307

The angular brackets denote integration with respect to coordinate x: h

K = h −1 ∫ Kdx 0

Owing to the symmetry of the swelling process in the layer, the problem is solved for 0<x< h. The equation of the liquid transport [6.1.3] in a plane layer has the form of a general diffusion equation except for the diffusion coefficient of the liquid, which, in the general case, is defined by the function D = D(N1,ν), implying that it depends on the liquid concentration and the relative longitudinal stretch of the layer.5 Eq. [6.1.4] is the law of conservation of matter for the polymeric matrix, and Eq. [6.1.5] states that the process of elastomer swelling is in the state of mechanical equilibrium. The initial condition is explicitly defined by Eq. [6.1.6]. The constraint that the diffusion flux and the displacements of polymeric matrix along x-axis in the layer center are absent is given by Eq. [6.1.7]. Eq. [6.1.8] has the physical meaning that there exists a thermodynamical equilibrium at the elastomer-liquid interface and that elastomer is not subjected to transverse mechanical loading, while Eq. [6.1.9] means that the layer does not experience longitudinal stretch under the external force. The assumption that the elastomer and the liquid are incompressible media can be mathematically represented by an incompressibility condition, which in the present case is written as5 J = φ−1

[6.1.10]

where the volume fraction of the polymer is φ = N2V2 / (N1V1 + N2V2 )

[6.1.11]

where: V1 and V2 the molar volumes of liquid and chains of the elastomer network, respectively

To make the definition of the examined problem complete, we need to add to the above model equations, the constitutive relations for mechanical stress tensor and chemical potential of a liquid. According to5, 6 these equation are given by

(

)

σ k = RTV2−1 λ k − I1 λ−k1 / 3 − pJλ−k1

[6.1.12]

µ = µ mix (φ) + RTZ −1 φ1/ 3 Γ1 / 3 + V1 p

[6.1.13]

[

µ mix = RT ln(1 − φ) + φ + χφ2

]

where: R T µ mix p χ

the gas constant per mole the absolute temperature the chemical potential of mixing pressure the Flory-Huggins interaction parameter

[6.1.14]

308


Z I1 Γ1

= V2/V1 = λ21 + λ22 + λ23 = I1/J2/3

The above equations follow from the classical high elasticity theory and the Flory theory of polymeric networks.7 From Eq. [6.1.5] and the second condition of Eq. [6.1.8], we find that σ1(x,t) = 0. This equation together with Eq. [6.1.12] yields the expression for pressure. By substituting it in the formulas for chemical potential [6.1.13] and longitudinal stresses, we find, using Eqs. [6.1.2] and [6.1.10], that

(

σ 2 = σ 3 = RTV2−1 ν − J 2 / ν 5 µ = µ mix (1 / J) + RTZ −1J / ν 4

)

[6.1.15] [6.1.16]

A substitution of Eq. [6.1.15] in Eq. [6.1.9] gives an expression for longitudinal stretch of the layer ν 6 = J 2 (x, t)

[6.1.17]

With consideration of Eq. [6.1.16], the boundary condition at x = h is transformed to µ mix (1 / J)RT + Z −1J / ν 4 = 0

[6.1.18]

Thus, the initial swelling problem for a plane layer is reduced to a boundary value problem for diffusion equation [6.1.3] with boundary conditions of Eqs. [6.1.6], [6.1.7], [6.1.17] and [6.1.18]. The solution to this problem provides a full description of swelling processes in the plane layer. In other words, using Eqs. [6.1.1], [6.1.2], [6.1.10] and [6.1.15] we can define a current distribution of a liquid through the layer and calculate the stress-strain state of the material. It should be noted that boundary conditions of Eq. [6.1.18] and Eq. [6.1.17] specify the existence of positive feedback in the system, which is responsible for the onset of unsteady boundary regime during material swelling. The nonlinear distributed systems with positive feedback are generally known as active media and are distinguished for their complex and multimode response.8 In free swelling, the response of elastomers is, in a sense, similar to that of active media. Such behavior is most pronounced when the extent of material swelling is high, which makes this case worthwhile for detailed investigation. For high-swelling elastomers, the volume fraction of polymer in equilibrium swelling state denoted in the following as ε and the volume fraction of polymer at the elastomer-liquid interface φ = 1/J entering Eq. [6.1.18] are small quantities. The asymptotic behavior of the function µmix(φ) at ϕ → 0 is described by µ mix (φ) / RT = −bφα

[6.1.19]

The constants b and α can be calculated using the Flory equation [6.1.14]. A second order expansion of ln(1-φ) as a power series of φgives b = 1/2-χ and α = 2. The scaling approach gives a slightly different value of α, which is found to be α = 9/4 (des Cloizeaux law9).


309

A volume fraction of the polymer in equilibrium swelling state can be determined by substituting Eq. [6.1.19] in Eq. 6.1.18] and setting φ = J-1 = ε and ν = ε-1/3, yields ε ≈ ( bZ ) −3/ ( 3α−1) . Then, using Eqs. [6.1.18], [6.1.19] and the last relation, we arrive at the following expression for the volumetric swelling ratio of the layer at the elastomer-liquid interface:

(

J ≈ ε −1 ε1/ 3 ν

)

6d

[6.1.20]

where d=

2 3(α + 1)

[6.1.21]

Note that approximate Eq. 6.1.20] defines the strain dependence of the equilibrium swell ratio of the elastomer in a liquid medium under conditions of biaxial symmetric material extension. Substituting Eq. [6.1.17] in Eq. [6.1.20], we express the boundary swell ratio in terms of liquid distribution in the layer J(h, t) = ε 2 d −1 J 2 (x, t)

d

[6.1.22]

Then the problem is finally defined as u t = (k(u, l )u x ) x ; x ∈ ( 01 , ), t > 0

[6.1.23]

u(x,0) = 0, u x (1, t) = 0

[6.1.24]

(1 − ε)u(0, t) + ε = [(1 − ε)u(x, t) + ε]

2

d

[6.1.25]

Here we assign dimensions to the variables. The quantities h and h2/D0 (where D0 is the value of diffusion coefficient in the state of ultimate elastomer swelling) are used as the units of distance and time. For the sake of convenience we transform, the coordinate to x→1-x. Integrating for x between the limits from 0 to 1 in Eq. [6.1.25] is designated by angular brackets. The function u(x,t) takes the value over the interval (0,1) and represents a dimensionless concentration of penetrating liquid. It is related to the liquid concentration and local material swelling by the following equations:

(

)

N1 = V1 ε −1 − 1 u(x, t), J(x, t) = ε −1 [(1 − ε)u(x, t) + ε]

[6.1.26]

The quantity l = ε1/ 3ν represents the longitudinal layer stretch normalized to unity. By virtue of [6.1.17] and [6.1.26] we may write l 6 (t) =

[(1− ε)u(x, t) + ε]

2

[6.1.27]

310


Dimensionless diffusion coefficient is defined by the formula k(u,l) = D(u,l)/D0. The longitudinal stresses in the layer (15) are expressed in terms of dimensionless stresses q(x,t) by σ 2 = σ 3 = RTV2−1q (x, t)

[6.1.28]

where according to Eqs. [6.1.15] and [6.1.27]

{

}

q (x, t) = ε −1/ 3 l (t) 1 − [(1 − ε)u(x, t) + ε] / l 6 (t) 2

[6.1.29]

Consider two functions g1 (t) = u(x, t) ,

g 2 (t) = u 2 (x, t)

[6.1.30]

which are integral characteristics of swelling kinetics for a plane layer and can be determined from experiments. The first function characterizes a relative amount of liquid absorbed by a polymer in time t and the second function according to Eq. [6.1.27] is related to longitudinal layer deformation. For high-swelling elastomers g 2 (t) ≈ l 6 (t)

[6.1.31]

The numerical results obtained by solving model problem of Eqs. [6.1.23] - [6.1.25] for a constant diffusion coefficient are plotted in Figure 6.1.1.6 The obtained curves show the evolution of penetrating liquid concentration and longitudinal stresses. It is seen that the boundary liquid concentration during swelling monotonically increases. 6.1.3 DIFFUSION KINETICS OF PLANE LAYER SWELLING Consider two stages of swelling process in a plane layer - the initial and final. In the initial stage, the influence of the opposite layer boundary on the swelling process is inessential and therefore diffusion in a layer of finite thickness at sufficiently small values of time can be considered as the diffusion in half-space.

Figure 6.1.1. Distribution of penetrating liquid (a) and longitudinal stresses (b) during swelling of a plane layer with constant diffusion coefficient k(u,l) = 1 at ε = 0.1 and d = 2/9: 1 - t = 0.05; 2 - t = 0.2; 3 - t = 0.4; 4 - t = 0.6; 5 - t = 1; 6 - t = 1.8. [Adapted, by permission, from E. Ya. Denisyuk, V. V. Tereshatov, Vysokomol. soed., A42, 74 (2000)].


311

At the very beginning of the swelling process the amount of absorbed liquid is rather small. Hence we may set u(x,t) ≈ 0 in the right-hand parts of Eqs. [6.1.25] and [6.1.27] which results in

(

)

u(0, t) ≈ ψ 0 = ε 2 d − ε / (1 − ε), l (t) ≈ ε1/ 3 and Eq. [6.1.23] becomes an usual parabolic equation describing diffusion on a half-line with constant boundary concentration ψ0. It has self-similar solution of the form u(x,t)=ψ0θ(x/t1/2). The function θ( ξ ) satisfies the equation (k(θ, ε1/ 3 )θ′)' + ξθ′/2 = 0 and the boundary conditions θ(0) = 1,

θ(+∞) = 0

[6.1.32]

From this follows the expression for the integral process characteristics g1 (t) = ψ 0 M1t 1/ 2 ,

g 2 (t) = ψ 20 M 2 t 1/ 2

where ∞

M p = ∫ θ p (ξ)dξ,

p = 1, 2

[6.1.33]

0

These relations define the asymptotic properties of swelling at t → 0. As more and more amount of the liquid is absorbed, the longitudinal strains in the layer increase. By virtue of Eq. [6.1.25], this causes the growth of liquid concentration at the boundary. For high-swelling materials at sufficiently large values of time, all terms in Eq. [6.1.25] involving ε as a multiplier factor can be neglected to the first approximation. The resulting expression is written as u(0, t) = u 2 (x, t)

d

[6.1.34]

where the angular brackets denote integrating for x in the limits from 0 to +∞. Since for arbitrary dependence of k(u,l) a boundary-value solution of equation [6.1.23] on the half-line with boundary condition of Eq. [6.1.34] cannot be represented in a similar form, we restrict our consideration to a model problem with diffusion coefficients defined by k(u, l ) = u s , s ≥ 0 k(u, l ) = u s l 6 p , s ≥ 0,

[6.1.35] p ≥0

[6.1.36]

(Let us agree that s = 0 corresponds to a constant diffusion coefficient k(u,l) = 1). The analysis of this problem allows us to qualitatively explain many mechanisms of diffusion kinetics of elastomer swelling. First, consider the diffusion coefficient defined by Eq. [6.1.35]. In this case, Eq. [6.1.23] on the half-line has a variety of self-similar solutions, which can be written as

312


u(x, t) = ψ(t)θ( x / ϕ(t))

[6.1.37]

where the function θ( ξ )satisfies conditions of Eq. [6.1.32] and defines the profile of the diffusion wave, ψ(t) describes boundary conditions and ϕ(t) the function specifies the penetration depth of diffusion wave. Eq. [6.1.37] satisfies the boundary condition of Eq. [6.1.34] and Eq. [6.1.23] on the half-line with the diffusion coefficient of Eq. [6.1.35] in the following cases: 1) ψ(t), ϕ(t) are the functions of power type (power swelling mode); 2) ψ(t), ϕ(t) are the function exponentially depending on time (exponential swelling mode); 3) ψ(t) ~ (t0 - t)m, ϕ(t)~(t0 - t)n, where m, n < 0 ( blow-up swelling mode). Power swelling mode occurs at sufficiently small values of s. In this case, the amount of absorbed liquid, boundary concentration, the depth of diffusion wave penetration and the longitudinal layer deformation are the power function of time. If the parameter s approaches the critical value sc = 2 / d − 4

[6.1.38]

Swelling process is governed by exponential law. And finally, if s > sc, the swelling mode is of a blow-up nature. The solutions describing these modes are given below. I. Power mode (s < sc): u(x, t) = M 22 m t m θ(ξ), g1 (t) = M1M 22 q −1t q , m=

ξ = x / M 2ms t n

[6.1.39]

g 2 (t) = M 22 r t r

[6.1.40]

1 1/ d − 2 1/ d − 1 1 , n= , q= , r= sc −s sc −s sc −s d (s c − s)

[6.1.41]

II. Exponential mode (s = sc):

{

}

{

}

u(x, t) = exp M 22 (t − t 0 ) θ(ξ), ξ = xM 2 / exp s c M 22 (t − t 0 ) / 2

{

}

g1 (t) = (M1 / M 2 ) exp (s c / 2 + 1)M 22 (t − t 0 )

{

}

g 2 (t) = exp (s c / 2 + 1)M 22 (t − t 0 ) III. Blow-up mode (s > sc): u(x, t) = M 22 m (t 0 − t) θ(ξ), m

g1 (t) = M1M 22 q −1 (t 0 − t) , q

ξ = x / M 2ms (t 0 − t) g 2 (t) = M 22 r (t 0 − t)

n

r

where the exponents are defined by Eqs. [6.1.41] but if s > sc, then m, n, q, rsc/2-1 the lower part of the kinetic curve is convex in a downward direction and the whole curve becomes S-shaped. Note that in terms of coordinates (t1/2,g1) at s ≥ 0 all the kinetic curves are S-shaped. Hence, the obtained solutions enable one to describe different anomalies of sorption kinetics observed in the experiments on elastomer swelling in low-molecular liquids. Figure 6.1.26 gives the results of numerical solution to problems of Eq. [6.1.25] with diffusion coefficient defined by Eq. [6.1.35]. The kinetic curves of swelling at different values of s are depicted in Figure 6.1.3.6

314


Figure 6.1.2. Diffusion kinetics of plane layer swelling for diffusion coefficient k(u)=us at ε = 0.1 and d = 2/9; a, b are power swelling modes at s = 1 and s = 2.5, respectively; c - exponential swelling mode (s=5); d - blow-up swelling mode (s = 5.5). Numerals over curves denote correspond to instants of time. [Adapted, by permission, from E. Ya. Denisyuk, V. V. Tereshatov, Vysokomol. soed., A42, 74 (2000)].

Here it is to be noted that strain dependence of the diffusion coefficient described by Eq. [6.1.36] does not initiate new diffusion modes. The obtained three self-similar solutions hold true. Only critical value sc is variable and is defined by expression sc = (2-p)/d-4. This fact can be supported by a direct check of the solutions. 6.1.4 EXPERIMENTAL STUDY OF ELASTOMER SWELLING KINETICS The obtained solutions can be applied to experimental study of the diffusive and thermodynamic properties of elastomers. In particular, with the relation s = 2/d - 4 - (1/d -1)/q

[6.1.45]


315

Figure 6.1.3. Kinetic curves of plane layer swelling at different values of concentration dependence of diffusion coefficient (ε = 0.1, d = 2/9): a - power law mode (1 - s = 0; 2 - s = 1; 3 - s = 1.5; 4 - s = 2.5); b - exponential (5 - s = 5) and blow-up mode (6 - s = 5.5). [Adapted, by permission, from E. Ya. Denisyuk, V. V. Tereshatov, Vysokomol. soed., A42, 74 (2000)].

following from Eqs. [6.1.38] and [6.1.41] we can estimate the concentration dependence of the diffusion coefficient of a liquid fraction in elastomer. According to Eq. [6.1.40] the parameter q is determined from the initial section of the kinetic swelling curve. Experimental estimates of the parameter r can be obtained from the strain curve l(t) using Eqs. [6.1.31] and [6.1.40]. Then by making use of the formula d = 1 - q/r

[6.1.46]

following from Eq. [6.1.41] we can evaluate the parameter d which characterizes the strain dependence of the equilibrium swelling ratio of elastomer under symmetric biaxial extension in Eq. [6.1.45]. Generally the estimation of this parameter in tests on equilibrium swelling of strained specimens proves to be a tedious experimental procedure. The value of diffusion coefficient in an equilibrium swelling state can be determined from the final section of kinetic swelling curve using Eq. [6.1.44], which is expressed in terms of dimensional variables as g1 (t ) = 1 − C exp( −α 12 D0 t / h 2 )

[6.1.47]

where α1 is calculated from Eq. [6.1.43). For d = 2/9, α1 ≈ 1.2220. Note that all these relations are valid only for sufficiently high values of elastomer swelling ratio. The obtained theoretical predictions have been verified in experiments on real elastomers. The elastomers tested in our experiments were amorphous polybutadiene urethanes (PBU) with polymer network of different density: 0.3 kmol/m3 (PBU-1), 0.05 kmol/m3 (PBU-2), 0.2 kmol/m3 (PBU-3), 0.1 kmol/m3 (PBU-4). Oligooxypropylene triol - Laprol 373 was used as a crosslinking agent at the curing of prepolymer of oligobutadiene diol. The elastomer specimens were manufactured in the form of disks, 35 mm in diameter and 2 mm thick. The kinetics of specimen swelling was determined in low-molecular liquids: toluene, dibutyl sebacate (DBS), dioctyl sebacate (DOS). The typical kinetic and strain curves of free swelling are given in Figure 6.1.4.6 The S-shape of the kinetic swelling curves in terms of coordinates (t1/2, g1) is indicative of

316


Figure 6.1.4. Kinetic (a) and strain (b) curves of elastomer swelling in toluene: 1 - PBU-3; 2 - PBU-4; 3 - PBU-1; 4 PBU-2. [Adapted, by permission, from E. Ya. Denisyuk, V. V. Tereshatov, Vysokomol. soed., A42, 74 (2000)].

abnormal sorption. The values of parameters q and r were obtained from kinetic and strain curves using the regression method. The values of correlation coefficient were 0.997 - 0.999 and 0.994 - 0.998 respectively. The obtained data and Eqs. [6.1.45], [6.1.46] were then used to calculate s and d. The diffusion coefficients were defined by the kinetic curves in terms of Eq. [6.1.47] under the assumption that d = 2/9. The obtained results were summarized in Table 6.1.1.6 The analysis of these data shows that swelling of the examined elastomers is of power-mode type. The concentration dependence of the liquid diffusion coefficient defined by the parameter s is found to be rather weak. For elastomers under consideration no exponential or blow-up swelling modes have been observed. Table 6.1.1. Experimental characteristics of elastomer swelling kinetics6 ε

q

r

s

d

D0, cm2/s

PBU-1/toluene

0.278

0.68

0.84

0

0.19

1.4×10-6

PBU-2/toluene

0.093

0.83

1.09

0.8

0.24

4.9×10-7

PBU-3/toluene

0.230

0.78

1.02

0.5

0.24

1.3×10-6

PBU-4/toluene

0.179

0.71

0.97

0

0.27

1.4×10-6

PBU-1/DBS

0.345

0.67

0.92

0

0.27

6.2×10-8

PBU-2/DBS

0.128

0.78

1.04

0.5

0.25

2.5×10-8

PBU-3/DBS

0.316

0.67

0.83

0

0.19

6.6×10-8

PBU-4/DBS

0.267

0.69

0.88

0

0.21

7.4×10-8

PBU-1/DOS

0.461

0.67

0.81

0

0.17

1.8×10-8

PBU-4/DOS

0.318

0.63

0.81

0

0.22

3.4×10-8

Elastomer/Liquid

It is of interest to note that experimental values of the parameter d characterizing the strain dependence of equilibrium swelling ratio for elastomers subjected to uniform biaxial extension closely approximate the theoretical values. It will be recalled that this parameter is specified by Eq. [6.1.21]. Moreover, the Flory theory defines it as d = 2/9 = 0.22(2),


317

whereas the des Cloizeaux law provides d ≈ 0,205, which suggests that the proposed model of elastomer swelling performs fairly well. 6.1.5 CONCLUSIONS In this section, we have developed a geometrically and physically nonlinear model of swelling processes for an infinite plane elastomeric layer and obtained approximate solutions describing different stages of swelling at large deformations of a polymeric matrix. We have identified the strain-stress state of the material caused by diffusion processes and analyzed its influence on the swelling kinetics. It has been found that a non-stationary boundary regime initiated by deformations arising in elastomer during swelling and increasing a thermodynamical compatibility of elastomer with a liquid is the main reason for swelling anomalies observed in the experiments. Anomalies of sorption kinetics turn out to be a typical phenomenon observable to one or another extent in elastic swelling materials. The theory predicts the possibility for qualitatively different diffusion modes of free swelling. A particular mode is specified by a complex of mechanical, thermodynamical, and diffusion material properties. The results of analytical and numerical solutions for a plane elastomer layer show that the swelling process may be governed by three different laws resulting in the power, exponential, and blow-up swelling modes. Experimentally it has been determined that in the examined elastomers the swelling mode is governed by the power law. The existence of exponential and blow-up swelling modes in real materials is still an open question. New methods have been proposed, which allow one to estimate the concentration dependence of liquid diffusion in elastomer and strain dependence of equilibrium swelling ratio under conditions of symmetric biaxial elastomer extension in terms of kinetic and strain curves of swelling. REFERENCES 1 2 3 4 5 6 7 8 9 10

A E Green, P M Naghdi, Int. J. Eng. Sci., 3, 231 (1965). A E Green, T R Steel, Int. J. Eng. Sci., 4, 483 (1966). K R Rajagopal, A S Wineman, MV Gandhi, Int. J. Eng. Sci., 24, 1453 (1986). M V Gandhi, K R Rajagopal, AS Wineman, Int. J. Eng. Sci., 25, 1441 (1987). E Ya Denisyuk, V V Tereshatov, Appl. Mech. Tech. Phys., 38, 913 (1997). E Ya Denisyuk, V V Tereshatov, Vysokomol. soed., A42, 74 (2000) (in Russian). P J Flory, Principles of polymer chemistry, Cornell Univ. Press, New York, 1953. V A Vasilyev, Yu M Romanovskiy, V G Yahno, Autowave Processes, Nauka, Moscow, 1987 (in Russian). P G De Gennes, Scaling Concepts in Polymer Physics, Cornell Univ. Press, Ithaca, 1980. A A Samarskiy, V A Galaktionov, S P Kurdyumov, A P Mikhaylov, Blow-up Modes in Problems for Quasilinear Parabolic Equations, Nauka, Moscow, 1987 (in Russian).

318

V V Tereshatov, V Yu. Senichev, E Ya Denisyuk

6.2 EQUILIBRIUM SWELLING IN BINARY SOLVENTS Vasiliy V. Tereshatov, Valery Yu. Senichev Institute of Technical Chemistry

E.Ya. Denisyuk Institute of Continuous Media Mechanics Ural Branch of Russian Academy of Sciences, Perm, Russia

Depending on the purposes and operating conditions of polymer material processing, the opposite demands to solubility of low-molecular-mass liquids in polymers exist. The products of polymer materials designed for use in contact with solvents should be stable against relative adsorption of these liquids. On the contrary, the well dissolving polymer solvents are necessary to produce the polymer films. The indispensable condition of creation of the plasticized polymer systems (for example, rubbers) is the high thermodynamic compatibility of plasticizers with the polymer basis of material. Hence, it immediately follows the statement of the problem of regulation of thermodynamic compatibility of polymers with low-molecular-mass liquids in a wide range of its concentration in polymer material. The problem of compatibility of crosslinked elastomers with mixed plasticizers and volatile solvents is thus of special interest. Depending on ratios between solubility parameters of the solvent 1, δ1, of the solvent 2, δ2, and polymer, δ3, solvents can be distinguished as “symmetric” liquids and “non-symmetric” ones. The non-symmetric liquids are defined as a mixture of two solvents of variable composition, solubility parameters δ1 and δ2 which are larger or smaller than solubility parameter of polymer (δ2 > δ1 > δ3, δ3 > δ2 > δ1). The symmetric liquid (SL) with relation to polymer is the mixture of two solvents, whose solubility parameter, δ1, is smaller, and parameter, δ2, is larger than the solubility parameter of polymer, δ3. The dependence of equilibrium swelling on the non-symmetric liquid composition does not have a maximum, as a rule.1 Research on swelling of crosslinked elastomers in SL is particularly interested in the regulation of thermodynamic compatibility of network polymers and binary liquids. Swelling in such liquids is characterized by the presence of maximum on the curve of dependence of network polymer equilibrium swelling and composition of a liquid phase.2,3 The extreme swelling of crosslinked polymers of different polarity in SL was discussed elsewhere.3 The following elastomers were used as samples: a crosslinked elastomer of ethylene-propylene rubber SCEPT-40 [δ3 = 16 (MJ/m3)1/2, (ve/V0)x = 0.24 kmol/m3], crosslinked polyester urethane, PEU, from copolymer of propylene oxide and trimethylol propane [δ3 = 18.3 (MJ/m3)1/2, (ve/V0)x = 0.27 kmol/m3], crosslinked polybutadiene urethane, PBU, from oligobutadiene diol [δ3 = 17.8 (MJ/m3)1/2, (ve/V0)x = 0.07 kmol/m3] and crosslinked elastomer of butadiene-nitrile rubber [δ3 = 19 (MJ/m3)1/2, (ve/V0)x = 0.05 kmol/m3]. The samples of crosslinked elastomers were swollen to equilibrium at 25oC in 11 SLs containing solvents of different polarity. The following regularities were established. With decrease in the solubility parameter value of component 1 (see Table 6.2.1) in SL (δ1 < δ3), the maximum value of equilibrium swelling, Q, shifts to the field of larger concentration of component 2 in the mixture (Figures 6.2.1 and 6.2.2). On the contrary, with decrease in the

6.2 Equilibrium swelling in binary solvents

Figure 6.2.1. Dependence of equilibrium swelling of PEU on the acetone concentration in the mixtures: 1-toluene-acetone, 2-cyclohexane-acetone, 3-heptane-acetone. [Adapted, by permission, from V. V. Tereshatov, M. I. Balashova, A. I. Gemuev, Prediction and regulating of properties of polymeric materials, Ural Branch of AS USSR Press, Sverdlovsk, 1989, p. 3.]

319

Figure 6.2.2. Dependence of equilibrium swelling of PBU on the DBP (component 2) concentration in the mixtures:1-DOS-DBP, 2-TO-DBP, 3- decane-DBP. [Adapted, by permission, from V. V. Tereshatov, M. I. Balashova, A. I. Gemuev, Prediction and regulating of properties of polymeric materials, Ural Branch of AS USSR Press, Sverdlovsk, 1989, p. 3.]

solubility parameter δ2 (see Table 6.2.1) of components 2 (δ2 > δ3), the maximum Q corresponds to composition of the liquid phase enriched by component 1 (Figure 6.2.3). Table 6.2.1. Characteristics of solvents and plasticizers at 298K. [Adapted, by permission, from V. V. Tereshatov, M. I. Balashova, A. I. Gemuev, Prediction and regulating of properties of polymeric materials, Ural Branch of AS USSR Press, Sverdlovsk, 1989, p. 3.] ρ, kg/m3

V×106, m3

δ, (MJ/m3)1/2

Cyclohexane

779

109

16.8

Heptane

684

147

15.2

Decane

730

194

15.8

Toluene

862

106

18.2

1,4-Dioxane

1034

86

20.5

Acetone

791

74

20.5

Ethyl acetate

901

98

18.6

Amyl acetate

938

148

17.3

Dibutyl phthalate

1045

266

19.0

Dioctyl sebacate

913

467

17.3

Transformer oil

890

296

16.0

Solvent/plasticizer

320


Figure 6.2.3. Dependence of equilibrium swelling of the crosslinked elastomer SCEPT-40 on the concentration of component 2 in the mixtures: 1-heptane-toluene, 2-heptane-amyl acetate, 3-heptane-ethyl acetate. [Adapted, by permission, from V. V. Tereshatov, M. I. Balashova, A. I. Gemuev, Prediction and regulating of properties of polymeric materials, Ural Branch of AS USSR Press, Sverdlovsk, 1989, p. 3.]

Figure 6.2.4. Dependence of equilibrium swelling of the crosslinked elastomer SCN-26 on the concentration of component 2 in the mixtures: 1-toluene-acetone, 2-ethyl acetate-dioxane, 3-ethyl acetate-acetone. [Adapted, by permission, from V. V. Tereshatov, M. I. Balashova, A. I. Gemuev, Prediction and regulating of properties of polymeric materials, Ural Branch of AS USSR Press, Sverdlovsk, 1989, p. 3.]

Neglecting the change of volume on mixing, the solubility parameter of the mixture of two liquids is represented by: δ12 = δ1 ϕ1 + δ 2 ϕ 2 where: ϕ 1 and ϕ 2 volume fractions of components 1 and 2

More exact evaluation of the δ12 value is possible if the experimental data on enthalpy of mixing, ∆H, of components of SL are taken into account:4

(

δ12 = δ12 ϕ1 + δ 22 ϕ 2 − ∆H / V12

)

1/ 2

With a change in δ1 and δ2 parameters, the SL composition has the maximum equilibrium swelling which corresponds to shifts in the field of composition of the liquid phase. The δ12 parameter is close or equal to the value of the solubility parameter of polymer. Such a simplified approach to the extreme swelling of polymers in liquid mixtures frequently works very well in practice. If there is a maximum on the curve of swelling in SL, then the swelling has an extreme character (11 cases out of 12) (Figures 6.2.1-6.2.4). The parameter of interaction, χ 123 , between polymer and a two-component liquid can be used as a co-solvency criterion for linear polymers (or criterion of extreme swelling), more general, than the equality (δ12 = δ3):5 χ 123 = χ 13 ϕ1 + χ 23 ϕ 2 − ϕ1 ϕ 2 χ 12

[6.2.1]


321

where: χ13 and χ 23 parameters of interaction of components 1 and 2 with polymer, correspondingly parameter of interaction of components 1and 2 of liquid mixture χ12

In the equation obtained from the fundamental work by Scott,5 mixed solvent is represented as “a uniform liquid” with the variable thermodynamic parameters depending on composition. If the χ123 value is considered as a criterion of existence of a maximum of equilibrium swelling of polymer in the mixed solvent, a maximum of Q should correspond to the minimum of χ123. For practical use of Eq. [6.2.2] it is necessary to know parameters χ13, χ23, and χ12. The values χ13 and χ23 can be determined by the Flory-Rehner equation, with data on swelling of a crosslinked elastomer in individual solvents 1 and 2. The evaluation of the χ12 value can be carried out with use of results of the experimental evaluation of vapor pressure, viscosity and other characteristics of a binary mixture.6,7 To raise the forecasting efficiency of prediction force of such criterion as χ123 minimum, the amount of performance parameters determined experimentally must be reduced. For this purpose, the following expression for the quality criterion of the mixed solvent (the analogy with expression of the Flory-Huggins parameter for individual solvent-polymer system) is used: s χ 123 = χ 123 +

V12 2 (δ3 − δ12 ) RT

V12 = V1 x 1 + V2 x 2

[6.2.2]

where: V12 V1, V2 x1, x2 R T s χ123

molar volume of the liquid mixture molar volumes of components 1 and 2 molar fractions of components 1 and 2 in the solvent universal gas constant temperature, K. constant, equal to 0.34

s The values of entropy components of parameters χ 13 and χ s23 (essential for quality pre8 diction) were taken into account. Using the real values of entropy components of interacs and χ s23 , we have: tion parameters χ 13

χ s13 = χ 13 −

V1 2 (δ3 − δ1 ) RT

[6.2.3]

χ s23 = χ 23 −

V2 2 (δ 3 − δ 2 ) RT

[6.2.4]

quality criterion of mixed solvent is represented by:3 s χ 123 = χ 13 ϕ1 + χ s23 ϕ 2 +

V12 2 (δ3 − δ12 ) RT

[6.2.5]

The values of parameters χ13 and χ23 were calculated from the Flory-Rehner equation, s , χ s23 , using data on swelling of crosslinked polymer in individual solvents, the values χ 13 and χ123 were estimated from Eqs. [6.2.3-6.2.5). Then equilibrium swelling of elastomer in SL was calculated from the equation similar to Flory-Rehner equation, considering the mix-

322


ture as a uniform liquid with parameters χ123 and V12 variable in composition. The calculated ratios of components of mixtures, at which the extreme swelling of elastomers is expected, are given in Table 6.2.2. Table 6.2.2. Position of a maximum on the swelling curve of crosslinked elastomers in binary mixtures. [Adapted, by permission, from V. V. Tereshatov, M. I. Balashova, A. I. Gemuev, Prediction and regulating of properties of polymeric materials, Ural Branch of AS USSR Press, Sverdlovsk, 1989, p. 3.]

Elastomer

PEU

PBU

Binary mixture

Calculation

Experiment

toluene-acetone

90/10

90/10

cyclohexanone-acetone

40/60

50/50

heptane-acetone

30/70

25/25

decane-DBP

30/70

30/70

TO-DBP

40/60

40/60

DOS-DBP

60/40

50/50

-

-

ethyl acetate-acetone

70/30

70/30

toluene-acetone

70/30

70/30

ethyl acetate-1,4-dioxane SCN-26

Components ratio, wt%

Results of calculation of the liquid phase composition at maximum swelling of elastomer correlate with the experimental data (see Table 6.2.2). The approach predicts the existence of an extremum on the swelling curve. This increases the forecasting efficiency of the prediction. The application of “approximation of the uniform liquid” (AUL) for prediction of extreme swelling of crosslinked elastomers is proven under condition of coincidence of composition of a two-component solvent in a liquid phase and in the swollen elastomer. Results of study of total and selective sorption by crosslinked elastomers of components of SLs are given below.3 Crosslinked PBU [(ve/V0) = 0.20 kmol/m3] and crosslinked elastomer of butadiene-nitrile rubber SCN-26 [(ve/V0) = 0.07 kmol/m3] were used in this study. The tests were carried out at 25±0.10oC in the following mixtures: n-nonane-tributyl phosphate (TBP), n-hexane-dibutyl phthalate (DBP), n-hexane-dibutyl maleate (DBM), and dioctyl sebacate (DOS)-diethyl phthalate (DEP). A crosslinked elastomer SCN-26 was immersed to equilibrate in amyl acetate-dimethyl phthalate mixture. Liquid phase composition was in range of 5 to 10%. A sol-fraction of samples (plates of 0.9×10-2 m diameter and in 0.3×10-2 m thickness) was preliminary extracted with toluene. For the high accuracy of the analysis of binary solvent composition, a volatile solvent, hexane, was used as a component of SL in most experiments. Following the attainment of equilibrium swelling, the samples were taken out of SL and held in air until the full evaporation of hexane, that was controlled by constant mass of sample. A content of a nonvolatile component of SL in elastomer was determined by the difference between the amount of liq-


Figure 6.2.5. Dependence of equilibrium swelling of PBU on the ϕ 2 value for DEP in the liquid phase (1) and the φ 2 value inside the gel (1') swollen in the mixture DOS-DEP.

323

Figure 6.2.6. Dependence of equilibrium swelling of PBU on the ϕ 2 value for DEP in the liquid phase (curves 1 and 2) and the φ 2 value inside the swollen gel (curves 1' and 2') in the mixtures: 1,1'-hexane(1)-DBM(2), 2,2'- hexane(1)-DBP(2).

uid in the swollen sample and the amount of hexane evaporated. In the case of SL with nonvolatile components (such as, DBP or DOS) the ratio of SL components in the swollen PBU was determined by the gas-liquid chromatography, for which purpose toluene extract was used. The volume fractions of components 1 and 2 of SL in a liquid phase ϕ1 and ϕ2 were calculated on the basis of their molar ratio and densities ρ 1 and ρ 2 . Volume fraction of polymer, υ3, in the swollen sample was calculated from the equation: υ3 =

1 QV + 1

where: QV

volume equilibrium swelling of elastomer in SL.

Volume fractions ϕ 1 and ϕ 2 of components 1 and 2 of low-molecular-mass liquids inside the swollen gel were determined from the equation:

φi =

υi , (i = 12 , ) 1− υ 3

where: φi

volume fraction of i-component related to the total volume of its low-molecular-mass part (not to the total volume of three-component system)

The results of study of sorption of two-component liquids in PBU and in crosslinked elastomer SCN-26 are shown in Figures 6.2.5-6.2.8, as dependencies of QV on the volume

324


Figure 6.2.7. Dependence of equilibrium swelling of PBU on the ϕ 2 value for TBP in the liquid phase (1) and the φ 2 value in the gel (1') swollen in the mixture: nonane-TBP.

Figure 6.2.8. Dependence of equilibrium swelling of crosslinked elastomer SCN-26 on the ϕ 2 value for DMP in the liquid phase ( 1) and the φ 2 value in the gel (1'), swollen in the mixture: amyl acetate-DMP.

fraction ϕ2 of component 2 in a liquid phase and on the volume fraction φ2 of component 2 of SL that is a part of the swollen gel. The data vividly show that extremum swelling of crosslinked elastomers in SL can be observed in all investigated cases. At the maximum value of QV, the compositions of SL in the liquid phase and in the swollen elastomer practically coincide (ϕ 2 ≈ φ2 ). The total sorption can be determined by the total content of the mixed solvent in the swollen skin9 or in the swollen elastic network (these results). The dependencies of QV on ϕ2 (Figures 6.2.5-6.2.8) reveal the Figure 6.2.9. Experimental dependence of the preferen- influence of liquid phase composition on tial sorption ε on the volume fraction ϕ 2 of component 2 the total sorption. The total sorption of the in the liquid phase: 1-DOS(1)-DEP(2)-PBU(3), 2-hex- binary solvent by polymer can also be meaane(1)-DBP(2)-PBU(3), 3-hexane(1)-DBM(2)-PBU(3), sured by the value of the volume fraction of 4-amyl acetate(1)-DMP(2)-SCN-26(3), 5-nonane(1) polymer in the swollen gel9 because the -TBP(2)-PBU(3). value of υ 3 is unequally related to the volume fraction of the absorbed liquid, υ 3 = 1− ( υ1 + υ 2 ). At the fixed total sorption (QV or υ 3 = const) the preferential sorption, ε, can be found from the following equation: ε = φ1 − ϕ1 = ϕ 2 − φ2


325

At the same values of Qv difference between coordinates on the abscissa axes of points of the curves 1-1', 2-2' (Figures 6.2.5-6.2.8) are equal ε (Figure 6.2.9). As expected,10 preferential sorption was observed, with the essential distinction of molar volumes V1 and V2 of components of the mixed solvent (hexane-DBP, DOS-DEP). For swelling of the crosslinked elastomer SCN-26 in the mixture of components, having similar molar volumes V1 and V2 (e.g., amyl acetate-dimethyl phthalate) the preferential sorption of components of SL is practically absent. Influence of V1 and V2 and the influence of double interaction parameters on the sorption of binary liquids by crosslinked elastomers was examined by the method of mathematical experiment. Therewith the set of equations describing swelling of crosslinked elastomers in binary mixture, similar to the equations obtained by Bristow6 from the Flory-Rehner theory11 and from the work of Schulz and Flory,12 were used: ln ϕ1 + (1 − l )ϕ 2 + χ 12 ϕ 22 = ln υ1 + (1 − l )υ 2 + υ 3 + χ 12 υ 22 + χ 12 υ 23 + 2   +(χ 12 + χ 13 − lχ 23 )υ 2 υ 3 + (ν e / V3 )V1  υ13/ 3 − υ 3  f  

(

)

(

[6.2.6]

)

ln ϕ 2 + 1 − l −1 ϕ1 + l −1 χ 12 ϕ12 = ln υ 2 + 1 − l −1 υ1 + υ 3 + l −1

χ12 υ1

2

2   + lχ 23 υ32 + ( χ12 + lχ 23 − χ13 ) υ1υ3  + (ν e / V3 )V2  υ31/3 − υ3  f  

[6.2.7]

where: l f

=V1/V2 functionality of a network

The analysis of calculations from Eqs. [6.2.6] and [6.2.7] has shown that if V1 χ13. The greater is χ12 value at V1 ≠ V2 the more likely preferential sorption takes place with all other parameters being equal. The same applies to the diluted polymer solutions. To improve calculations of the preferential sorption in the diluted solutions of polymers, the correction of Flory’s theory is given in works9,13-15 by introduction of the parameter of three-component interaction, χT, into the expression for free energy of mixing, and substitution of χT by qT.14 This approach is an essential advancement in the analysis of a sorption of two-component liquids by polymer. On the other hand, the increase in the number of experimental parameters complicates the task of prediction of the preferential sorption. In swelling crosslinked elastomers, the preferential sorption corresponds to the maximum QV value and compositions in a swollen polymer and a liquid phase practically coincide (Figures 6.2.5-6.2.8). This observation can be successfully used for an approximate evaluation of the preferential sorption. Assuming that ϕ1 = φ1 and ϕ 2 = φ2 at extremum of the total sorption (φ3 = φ3min ), AUL5 e can be used to calculate the effective value of χ 12 from the expression for χ123:16 e = χ 12

χ 13 ϕ 2m

+

χ 23V1 ϕ1mV2

−

m V1 χ 123

V12m ϕ1m ϕ 2m

[6.2.8]

326

Figure 6.2.10. Experimental data and calculated dependence of equilibrium fraction of components υ1 (1), υ2 (2), and υ3 (3) in PBU sample swollen in the hexane-DBP mixture on the ϕ 2 value for DBP in the liquid phase: lines - calculation, points - experimental.


Figure 6.2.11. Dependence of equilibrium fraction of components υ1 (1), υ2 (2), and υ3 (3) in SCN-26 sample swollen in the amyl acetate-DMP mixture on the ϕ 2 value for DMP in the liquid phase: lines - calculation, points - experimental. where: m χ12e

the index for the maximum of sorption. the value, calculated from the equation [6.2.8], substitutes χ12 from the Eqs. [6.2.6] and [6.2.7]

The results of calculations are presented as dependencies of volume fractions, υ1 , υ 2 , and υ 3 of the triple system components (swollen polymer) vs. the volume fraction ϕ 2 (or ϕ1) of the corresponding components of the liquid phase (Figures 6.2.10 and 6.2.11). For practical purposes it is convenient to represent the preferential sorption as the dependence of equilibrium composition of a binary Figure 6.2.12. Dependence of equilibrium φ 2 value on liquid (a part of the swollen gel) vs. composithe ϕ 2 value in the liquid phase: 1-amyl acetate (1)-DMP (2)-SCN-26(3), 2-hexane(1)-DBM(2)-PBU tion of the liquid phase. These calculated de(3), 3-nonane(1)-TBP(2)- PBU (3); lines - calculation, pendencies (solid lines) and experimental points - experimental. (points) data for three systems are given in Figure 6.2.12. The results of calculations based on AUL, are in the satisfactory agreement with experimental data. Hence the experimentally determined equality of concentrations of SL components in the swollen gel and in the liquid phase allows one to predict composition of liquid, whereby polymer swelling is at its maximum, and the preferential sorption of components of SL. To refine dependencies of φ1 on ϕ1 or φ2 on ϕ 2 , correction can be used, for example, minimization of the square-law deviation of calculated and experimental data on the total

6.3 Swelling data

327

sorption of SL by polymer. Thus it is necessary to account for proximity of the compositions of solvent in the swollen gel and the liquid phase. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

A Horta, Makromol. Chem., 182, 1705 (1981). V V Tereshatov, V Yu Senichev, A I Gemuev, Vysokomol. Soedin., 32B, 422 (1990). V V Tereshatov, M I Balashova, A I Gemuev, Prediction and regulating of properties of polymeric materials, Ural Branch. of AS USSR Press, Sverdlovsk, 1989, p. 3. A A Askadsky, Yu I Matveev, Chemical structure and physical properties of polymers, Chemistry, Moscow, 1983. R L Scott, J. Chem. Phys., 17, 268 (1949). C M Bristow, Trans. Faraday Soc., 55, 1246 (1959). A Dondos, D Patterson, J. Polym. Sci., A-2, 5, 230 (1967). L N Mizerovsky, L N Vasnyatskaya, G M Smurova, Vysokomol. Soedin., 29A, 1512 (1987). J Pouchly, A Zivny, J. Polym. Sci., 10, 1481 (1972). W R Krigbaum, D K Carpenter, J. Polym. Sci., 14, 241 (1954). P J Flory, J Rehner, J. Chem. Phys., 11, 521 (1943). A R Shulz, P J Flory, J. Polym. Sci., 15, 231 (1955). J Pouchly, A Zivny, J. Polym. Sci., 23, 245 (1968). J Pouchly, A Zivny, Makromol. Chem., 183, 3019 (1982). R M Msegosa, M R Comez-Anton, A Horta, Makromol. Chem., 187, 163 (1986). J Scanlan, J. Appl. Polym. Sci., 9, 241 (1965).

6.3 SWELLING DATA ON CROSSLINKED POLYMERS IN SOLVENTS

Vasiliy V. Tereshatov, Valery Yu. Senichev Institute of Technical Chemistry Ural Branch of Russian Academy of Sciences, Perm, Russia

Data on equilibrium swelling of selected crosslinked elastomers in different solvents are presented in Table 6.3.2. All data were obtained in the author’s laboratory during the last 5 years. The network density values for these elastomers are given in Table 6.3.1. The data can be used for the quantitative evaluation of thermodynamic compatibility of solvents with elastomers and for calculation of ∆Z, the thermodynamic interaction parameter. These data can be recalculated to the network density distinguished by the network density value, given in Table 6.3.1, using Eq. [4.2.9]. Network density values can be used for the calculation of the interaction parameter. It should be noted that each rubber has specific ratio for the equilibrium swelling values in various solvents. This can help to identify rubber in polymeric material. Temperature influences swelling (see Subchapter 4.2). Equilibrium swelling data at four different temperatures are given in Table 6.3.3. Polymer samples were made from industrial rubbers. The following polyols were used for preparation of polyurethanes: butadiene diol (M~2000) for PBU, polyoxybutylene glycol (M~1000) for SCU-PFL, polyoxypropylene triol (M~5000) for Laprol 5003, polydiethylene glycol adipate (M~2000) for P-9A and polydiethylene glycol adipate (M~800) for PDA. Polyurethanes from these polyols were synthesized by reaction with 2,4-toluylene diisocyanate with addition of crosslinking agent - trimetylol propane, PDUE, synthesized by reaction of oligobutadiene isoprene diol (M~4500) with double excess of 2,4-toluylene diisocyanate, followed by the reaction with glycidol.

328

Vasiliy V. Tereshatov, Valery Yu. Senichev

Table 6.3.1. Network density values for tested elastomers (data obtained from the elasticity modulus of samples swollen in good solvents) Elastomer

Network density, (kmol/m3)×102

Trade mark

Silicone rubber

SCT

6.0

Butyl rubber

BR

8.9

Polybutadiene rubber

SCDL

40.5


SCEPT

2.7

Isoprene rubber

SCI-NL

8.9


SCN-26

5.1

Polydiene-urethane-epoxide

PDUE

16.0

Polybutadiene urethane

PBU

20.8

Polyoxybutylene glycol urethane

PFU

10.0

Polyoxypropylene glycol urethane

Laprol 5003

43.0

Polydiethylene glycol adipate urethane

P-9A

6.4

Polydiethylene glycol adipate urethane

PDA

9.6

Table 6.3.2 Equilibrium swelling data (wt%) at 25oC

Solvent\elastomer

SCT

BR

SCDL

SCEPT

SCI

SCN

-NL

-26

PDUE

PBU

122

78

137

84

SCU

Laprol

P-9A

16

44

4

16

42

0

101

32

0

7

139

28

1

155

315

39

24

238

3

53

4

0.5

-PFL

PDA

Hydrocarbons Pentane

355

75

Hexane

380

305

Heptane

367

357

Isooctane

380

341

Decane

391

Cyclohexane

677

Transformer oil

47

105

214 586

300

881

10

1020

230

1580

600

3

Toluene

384

1294

435

492

356

134

315

77

31

Benzene

233

511

506

479

359

158

340

175

73

o-Xylene

515

496

356

122

288

53

42

87

2

Ethers Diamyl

310

220

889

584

20

248

6.3 Swelling data

329

Solvent\elastomer

SCT

Diisoamyl

333

BR

SCDL

SCEPT

SCI

SCN

-NL

-26

791 120

Diethyl of diethylene glycol

85

428

180

896

156

702

7

168

11

203

SCU -PFL

6

242

Laprol

P-9A

Tetrahydrofuran

PDA

3 22

2

41

0.1

251

23

1,4-Dioxane

PBU

33

Didecyl Dioctyl

PDUE

1 54

264

178

782

48

192

848

133

213

88

121

64

40

238

48

35

249

36

Esters of monoacids Heptyl propionate

260

236

Ethyl acetate

20

Butyl acetate

55

Isobutyl acetate

304

382

287 170

55

562

317

251

48

267

210

Amyl acetate

82

343

Isoamyl acetate

74 213

Heptyl acetate

236

100

22

545

100

6

Methyl capronate

369

285

115

Isobutyl isobuturate

328

276

64

260

36

26

6

6

Esters of multifunctional acids Dihexyl oxalate

28

350

Diethyl phthalate

6

837

47

Dibutyl phthalate

9

767

123

73

166

11

9

93

165

272

496

126

154

121

228

47

20

187

9

570

Dihexyl phthalate Diheptyl phthalate Dioctyl phthalate

8

28

189 211

Dinonyl phthalate

181

Didecyl phthalate Diamyl maleate

425

Dimethyl adipate

560 294

Diamyl adipate

3

29

1

33

1.4

23

0.5

71

203

20

19

14

246

716

6

8 295

Dioctyl adipate

232

88

280

Dihexyl adipate

Dipropyl sebacate

197

37

199

42

2

2

126

12

0.6

330

Solvent\elastomer


SCT

BR

SCDL

SCEPT

SCI

SCN

-NL

-26

PDUE

SCU

PBU

-PFL

P-9A

70

Diamyl sebacate

60

Diheptyl sebacate Dioctyl sebacate

Laprol

13

160

127

445

39

Triacetyne

34

Tricresyl phosphate

800

261

5.5

15

28

45 300

Tributyl phosphate

162

2 118

3

74

0.4

0.7

489

129

140 244

238

PDA

80 256

93

327

523

105

Ketones 577

Cyclohexanone

7

Acetone

44

88

218

99

Alcohols 49

Ethanol Butanol

16

25

72

7

12

Pentanol

28

33

66

10

5

Hexanol

46

34

62

9

4

36

67

19

Heptanol

57

Nonanol

3

Halogen compounds 2830

CCl4

284

999

584 234

Chlorobenzene

1565

Chloroform Fluorobenzene

66 277

114

572

616

205

113

120

34

Nitrogen compounds 336

Diethyl aniline

52

Dimethylformamide

277

Nitrobenzene Capronitrile

769

142

Acetonitrile

47

69

918

298

242 115

16

799

241

84 323

317

Aniline

377

423

Table 6.3.3 Equilibrium swelling data (wt%) at -35,-10, 25, and 50oC Solvent\Elastomer Isopropanol

Laprol* 8,12,37,305

SCU-PFL 8,19,72,-

P-9A 8,10,18,28

PBU 3,5,18,28

SCN-26 20,23,26,41

6.4 Influence of structure on equilibrium swelling

Solvent\Elastomer

Laprol*

SCU-PFL

331

P-9A

PBU

SCN-26

Pentanol

27,34,80,119

16,25,78,104

7,9,10,18

8,9,33,52

31,37,49,60

Acetone

80,366,381,-

80,122,132,-

48,73,218,-

56,321,347,-

162,254,269,382

Ethyl acetate

214,230,288,314

113,114,133,138

209,210,210,213

165,171,187,257

324,343,395,407

Butyl acetate

294,304,309,469

119,122,123,125

41,44,64,80

288,293,307,473

404,426,518,536

Isobutyl acetate

259,259,288,-

89,92,100,105

23,28,48,57

210,212,223,348

309,326,357,362

Amyl acetate

315,309,315,426

124,136,144,155

16,18,36,44

285,297,298,364

343,347,358,359

Tetrahydrofuran

91,96,97,119

119,177,192,-

838,840,848,859

45,51,57,80

145,150,166,-

o-Xylene

363,368,389,471

117,121,122,137

24,25,53,60

328,357,369,375

428,440,405,406

Chlorobenzene

522,537,562,-

241,242,242,246

238,244,277,333

517,528,559,698

967,973,985,1256

Acetonitrile

19,21,28,33

23,26,39,41

254,263,323,362

6,10,16,27

47,47,47,82

Hexane

50,52,61,75

2,3,9,14

4,4,4,28

51,56,63,80

-

120,122,134,149

50,57,77,85

331,339,372,415

-

Toluene

360,366,380, 397

*(ve/V) = 0.149 kmol/m3. Network density values for other elastomers correspond to Table 6.3.2

6.4 INFLUENCE OF STRUCTURE ON EQUILIBRIUM SWELLING Vasiliy V. Tereshatov, Valery Yu. Senichev Institute of Technical Chemistry Ural Branch of Russian Academy of Sciences, Perm, Russia

Swelling of single-phase elastomers is, other parameters being equal, limited by the chemical network. Microphase separation of hard and soft blocks can essentially influence swelling of block-copolymers. Hard domains formed in this process are knots of physical network, which can be resistant to action of solvents.1-4 Swelling of polyurethane block-copolymers with urethane-urea hard segments is investigated. The maximum values of swelling, Qmax, of segmented polyurethane (prepolymer of oligopropylene diol with functional isocyanate groups) cured by 4,4'-methylene-bis-o-chloroaniline (MOCA) are given in Tables 6.4.1 and 6.4.2. The prepolymer Vibratane B 600 was obtained by the reaction of oligopropylene diol with 2,4-toluylene diisocyanate. The swelling experiments were carried out in four groups of solvents of different polarity and chemical structure.5 The effective molecular mass of elastically active chains, Mc, between network crosslinks was estimated from the Flory-Rehner equation.6 The interaction parameter of solvent with polymer was determined by calculation. In the first variant of calculation, a classical method based on the solubility parameter concept, was used with application of Bristow and Watson’s semi-empirical relationship for χ 1:7

332

Vasiliy V. Tereshatov, Valery Yu. Senichev 2 V  χ 1 = χ S1 +  1  ( δ1 − δ p )  RT 

[6.4.1]

where: χ1S V1 R1 T δ1 and δp

a lattice constant whose value can be taken as 0.34 the molar volume of the solvent the gas constant the absolute temperature the solubility parameters of the solvent and the polymer, respectively

However, to evaluate χ 1 from Eq. [6.4.1], we need accurate data for δ1. In some cases the negative Mc values are obtained (Tables 6.4.1, 6.4.2). The reason is that the real values of the entropy component of the χ 1 parameter can strongly differ from the 0.34 value. Table 6.4.1. Characteristics of polyurethane-solvent systems at 25oC. [Adapted, by permission, from U.S. Aithal, T.M. Aminabhavi, R.H. Balundgi, and S.S. Shukla, JMS Rev. Macromol. Chem. Phys., 30C (1), 43 (1990).] Penetrant

Qmax

χ1

MC

Monocyclic aromatics Benzene

0.71

0.378

630

Toluene

0.602

0.454

790

p-Xylene

0.497

0.510

815

Mesitylene

0.402

0.531

704

Chlorobenzene

1.055

0.347

905

Bromobenzene

1.475

0.347

1248

o-Dichlorobenzene

1.314

0.357

1106

Anisole

0.803

0.367

821

Nitrobenzene

1.063

0.356

835

Methanol

0.249

1.925

-90

Ethanol

0.334

0.349

172

n-Propanol

0.380

0.421

484

Isopropanol

0.238

0.352

249

n-Butanol

0.473

2.702

-131

2-Butanol

0.333

1.652

-307

2-Methyl-1-propanol

0.389

0.440

535

Isoamyl alcohol

0.414

0.357

398

Aliphatic alcohols


333

Table 6.4.2. Characteristics of polyurethane-solvent systems at 25oC. [Adapted, by permission, from U.S. Aithal, T.M. Aminabhavi, R.H. Balundgi, and S.S. Shukla, JMS Rev. Macromol. Chem. Phys., 30C (1), 43 (1990).] Penetrant

Qmax

χ1

MC

Halogenated aliphatics Chloroform

4.206

0.362

3839

Bromoform

5.583

0.435

4694

1,2-Dibromoethane

1.855

0.412

770

1,3-Dibromopropane

1.552

0.444

870

Dichloromethane

2.104

0.34

1179

Trichloroethylene

1.696

0.364

1098

Tetrachlorethylene

0.832

0.368

429

1,2-Dichlorethane

1.273

0.345

728


1.058

0.538

860

1,4-Dichlorobutane

0.775

0.348

626

1,1,2,2- Tetrachloroethane

5.214

0.34

6179

Miscellaneous liquids Methyl acetate

0.494

0.341

370

Ethyl acetate

0.509

0.422

513

Ethyl benzoate

0.907

0.34

1111

Methyl ethyl ketone

1.261

0.364

2088

Tetrahydrofuran

2.915

0.39

2890

1,4-Dioxane

2.267

0.376

2966

DMF

1.687

1.093

-1184

DMSO

0.890

1.034

-975

Acetonitrile

0.184

0.772

182

Nitromethane

0.302

1.160

5734

Nitroethane

0.463

0.578

382

n-Hexane

0.069

1.62

2192

Cyclohexane

0.176

0.753

287

Benzyl alcohol

4.221

In the second variant of evaluation of Mc, data on the temperature dependence of volume fraction of polymer, ϕ 2 , in the swollen gel were used. Results of calculations of Mc from the Flory-Rehner equation in some cases also gave negative values. The evaluation of

334


Mc in the framework of this approach is not an independent way, and Mc is an adjustment parameter, as is the parameter χ 1 of interaction between solvent and polymer. In the last decades of evaluation of the physical network density of SPU was done on samples swollen to equilibrium in two solvents.2,9,10 Swelling of SPU in toluene practically does not affect hard domains.9,11 Swelling of SPU (based on oligoethers diol) in a tributyl phosphate (a strong acceptor of protons) results in full destruction of the physical network with hard domains.2 The effective network density was evaluated for samples swollen to equilibrium in toluene according to the Cluff-Gladding method.12 Samples were swollen to equilibrium in TBP and the density of the physical network was determined from equation:2,10

(v e

/ V0 ) dx − (v e / V0 ) x = (v e / V0 ) d

[6.4.2]

As the result of unequal influence of solvents on the physical network of SPU, the values of effective density of networks calculated for the “dry” cut sample can essentially differ. The examples of such influence of solvents are given in the work.2 SPU samples with oligodiene soft segments and various concentration of urethane-urea hard blocks were swollen to equilibrium in toluene, methyl ethyl ketone (MEK), tetrahydrofuran (THF), 1,4-dioxane and TBP (experiments 1-6, 9, 10, 12). SPU based on oligoether (experiment 7), with urethane-urea hard segments, and crosslinked single-phase polyurethanes (PU) on the base of oligodiene prepolymer with functional isocyanate groups, cured by oligoether triols (experiments 8, 11) were also used. The effective network densities of materials swollen in these solvents were evaluated by the Cluff-Gladding method through the elasticity equilibrium modulus. The data are given in Table 6.4.3 per unit of initial volume. The data shows that the lowest values of the network density are obtained for samples swollen to equilibrium in TBP. Only TBP completely breaks down domains of hard blocks. If solvents which are acceptors of protons (MEK, THF, 1,4-dioxane) are used in swelling experiment an intermediate values are obtained between those for toluene and TBP. The network densities of SPU (experiments 8 and 11) obtained for samples swollen in toluene, TBP, 1,4-dioxane and THF coincide (Table 6.4.3).2 Table 6.4.3. Results of the evaluation of equilibrium swelling and network parameters of PUE. [Adapted, by permission, from E.N. Tereshatova, V.V. Tereshatov, V.P. Begishev, and M.A. Makarova, Vysokomol . Soed., 34B, 22 (1992).] #

ρ kg/m3

QV

1

996

1.56

1.03

0.46

0.77

1.08

0.41

3.76

0.12

5.12

0.06

2

986

2.09

0.53

0.52

0.43

1.40

0.23

6.22

0.04

11.15

0.02

3

999

1.74

0.77

0.49

0.73

1.11

0.38

7.88

0.03

18.48

0.01

4

1001

2.41

0.44

0.49

1.75

0.13

14.3

0.02

∞

5

1003

4.46

0.15

0.47

3.29

6

984

2.63

0.40

0.57

7

1140

0.86

1.83

2.06

(ve/V0)

Toluene

QV

(ve/V0)

THF

0.39

QV

(ve/V0)

MEK

QV

(ve/V0)

QV

1,4-Dioxane

∞

(ve/V0)

TBP

∞

1.45

0.19

3.44

0.14

4.21

0.08

1.40

0.73

1.86

0.31

7.22

0.04


335

#

ρ kg/m3

QV

8

972

2.42

0.35

1.84

0.36

1.01

0.54

2.08

0.34

1.48

0.35

9

979

2.26

0.39

1.72

0.37

0.98

0.54

1.97

0.35

1.62

0.31

10

990

2.00

0.63

0.47

0.99

0.55

2.42

0.29

2.19

0.24

11

991

4.11

0.18

2.79

0.18

3.98

0.17

3.42

0.18

2.54

0.17

12

997

2.42

0.50

1.00

0.32

2.68

0.12

4.06

0.13

4.32

0.08

(ve/V0)

Toluene

QV

(ve/V0)

THF

QV

(ve/V0)

MEK

QV

(ve/V0)

1,4-Dioxane

QV

(ve/V0)

TBP

(ve/V0), kmol/m3

To understand the restrictions to swelling of SPU caused by the physical network containing hard domains, the following experiments were carried out. Segmented polybutadiene urethane urea (PBUU) on the base of oligobutadiene diol urethane prepolymer with functional NCO-groups (M ≈ 2400), cured with MOCA, and SPU-10 based on prepolymer cured with the mixture of MOCA and oligopropylene triol (M ≈5000) were used. The chemical network densities of PBUU and SPU were 0.05 and 0.08 kmol/m3, respectively. The physical network density of initial sample, (ve/V0)d, of PBUU was 0.99 kmol/m3 and of SPU-10 was 0.43 kmol/m3. Samples of PBUU and SPU-10 were swollen to equilibrium in solvents of different polarity: dioctyl sebacate (DOS), dioctyl adipate (DOA), dihexyl phthalate (DHP), transformer oil (TM), nitrile of oleic acid (NOA), dibutyl carbitol formal (DBCF), and tributyl phosphate (TBP). The values of equilibrium swelling of elastomers in these solvents, Q1, (ratio of the solvent mass to the mass of the unswollen sample) are given in Table 6.4.4. After swelling in a given solvent, samples were swollen in toluene to equilibrium. The obtained data for swelling in toluene, QVT , indicate that the physical networks of PBUU and SPU-10 do not change on swelling in most solvents. Equilibrium swelling in toluene of initial sample and the sample previously swollen in other solvents is practically identical. Several other observations were made from swelling experiments, including sequential application of different solvents. If preliminary disruption of the physical network of PBUU and SPU-10 by TBP occurs, swelling of these materials in toluene strongly increases. Similarly, samples previously swollen in TBP have higher equilibrium swelling, Q2, when swollen in other solvents. The value of Q2 is likely higher than equilibrium swelling Q1 of initial sample (Table 6.4.4). Q2 for PBUU is closer to the value of equilibrium swelling of a single-phase crosslinked polybutadiene urethane, PBU, having chemical network density, (ve/V0)x = 0.04 kmol/m3. Thus, the dense physical network of polyurethane essentially limits the extent of equilibrium swelling in solvents, which do not breakdown the domain structure of a material.

336


Table 6.4.4. Equilibrium swelling of SPBUU and SPU-10 (the initial sample and sample after breakdown of hard domains by TBP) and swelling of amorphous PBU at 25oC SPBUU Solvent

Initial structure

Q1 -

QT

SPU-10

After breakdown of domains

Q2

Initial structure

Q1

1.35

PBU

After breakdown of domains

QT

Q2

Q3

5.56

2.10

DOS

0.62

1.34

3.97

0.66

2.09

3.04

DOA

0.68

1.34

4.18

0.84

2.05

3.26

DBP

0.60

1.36

3.24

0.90

2.12

3.38

DHP

0.63

1.34

3.71

0.85

2.10

3.22

4.47

TO

0.57

1.35

2.29

0.58

2.07

1.92

1.99

NOA

0.59

1.36

4.03

0.72

2.10

2.86

4.66

DBCP

0.69

1.37

4.60

0.155

2.96

3.45

5.08

5.01 7.92*

5.01

4.22

5.43

4.22

TBP

* The result refers to the elastomer, having physical network completely disrupted

Swelling of SPU can be influenced by changes in elastomer structure resulting from mechanical action. At higher tensile strains of SPU, a successive breakdown of hard domains as well as micro-segregation may come into play causing reorganization.13-15 If the structural changes in segmented elastomers are accompanied by breakdown of hard domains and a concomitant transformation of a certain amount of hard segments into a soft polymeric matrix or pulling of some soft blocks (structural defects) out of the hard domains,14 variation of network parameters, is inevitable. It is known4 that high strains applied to SPU causes disruption of the physical network and a significant drop in its density. The data of equilibrium swelling of SPU-1 with oligodiene soft segments obtained by curing prepolymer by the mixture of MOCA and oligobutadiene diol (M ≈ 2000) with the molar ratio 1/1 are given in Table 6.4.5. The initial density of the SPU-1 network (ve/V0)dx = 0.206 kmol/m3, the physical network density (ve/V0)d = 0.170 kmol/m3. After stretching by 700 % and subsequent unloading, the value (ve/V0)dx = 0.115 kmol/m3 and (ve/V0)d = 0.079 kmol/m3. The density of the chemical network (ve/V0)x = 0.036 kmol/m3 did not change. Samples were swollen to equilibrium in a set of solvents: toluene, n-octane, cyclohexane, p-xylene, butyl acetate, DBCF, DBP, DHP, dihexyl sebacate (DHS), DOA and DOS (the values of density, ρ1, and molar volume of solvents, V1, are given in Table 6.2.5). The data in Table 6.4.5 shows that, after stretching, the volume equilibrium swelling of samples does not change in DBCF.4 In other solvents equilibrium swelling is noticeably increased. Swelling experiments of samples previously swollen in solvents and subsequently in toluene have shown that the value of equilibrium swelling in toluene varies only for samples previously swollen in DBCF. The effective network densities evaluated for SPU-1 samples, swollen in DBCF, and then in toluene, have appeared equal (0.036 and 0.037 kmol/m3, respectively) and these values correspond to the value of the chemical network parameter,


337

(ve/V0)x, of SPU-1. It means that DBCF breaks down the physical network of elastomer; thus, the values of Qv in DBCF do not depend on the tensile strain. In all other cases, swelling of SPU-1 in toluene differs only marginally from the initial swelling of respective samples in toluene alone. Therefore, these solvents are unable to cause the breakdown of the physical network. Table 6.4.5. Equilibrium swelling of SPU-1 samples in various solvents before and after stretching at 25oC (calculated and experimental data). [Adapted, by permission, from V.V. Tereshatov, Vysokomol. soed., 37A, 1529 (1995).] 3

ρ 1, 3 kg/m

V1×10 , 3 m /kmol

QV (ε=0)

χ1 (ε=0)

Toluene

862

107

3.54

DBP

1043

266

n-Octane

698

Cyclohexane

774

p-Xylene

Solvent

QV (ε=700%)

QTV (ε=700%)

QTV (ε=0)

calc

exp

0.32

5.06

4.87

1.12

0.56

1.47

1.53

4.85

3.49

164

0.79

0.75

0.90

0.93

4.94

3.60

109

1.46

0.60

1.78

1.95

4.99

3.63

858

124

2.94

0.36

4.17

3.93

4.81

3.52

Butyl acetate

876

133

2.28

0.44

3.13

2.97

5.12

3.69

DOS

912

468

1.21

0.38

1.60

1.71

5.10

3.63

DOA

924

402

1.49

0.32

2.19

2.18

5.00

3.61

DHS

923

340

1.74

0.30

2.55

2.44

5.01

3.59

DHP

1001

334

1.25

0.47

1.71

1.74

4.79

3.47

DBCP

976

347

4.48

4.51

9.92

9.98

The calculation of equilibrium swelling of SPU-1 (after its deformation up to 700 % and subsequent unloading) in a given solvent was carried out using the Flory - Rehner equation. In our calculations we used values of the χ 1 parameter of interaction between solvents and elastomer calculated from experimental data of equilibrium swelling of initial sample (ε=0) in these solvents. The diference between calculated and experimental data on equilibrium swelling of SPU-1 samples after their deformation does not exceed 9%. Therefore, the increase in swelling of SPU-1 is not related to the change of the χ 1 parameter but it is related to a decrease in the (ve/V0)dx value of its three-dimensional network, caused by restructure of material by effect of strain. The change of the domain structure of SPU exposed to increased temperature (~200oC) and consequent storage of an elastomer at room temperature may also cause a change in equilibrium swelling of material.16 REFERENCES 1 2 3

V.P. Begishev, V.V. Tereshatov, E.N. Tereshatova, Int. Conf. Polymers in extreme environments, Nottingham, July 9-10, 1991, Univ. Press, Nottingham, 1991, pp. 1-6. E.N. Tereshatova, V.V. Tereshatov, V.P. Begishev, and M.A. Makarova, Vysokomol. Soed., 34B, 22 (1992). V.V. Tereshatov, E.N. Tereshatova, V.P. Begishev, V.I. Karmanov, and I.V. Baranets, Polym. Sci., 36A, 1680 (1994).

338

4 5 6 7 8 9 10 11 12 13 14 15 16

Vasiliy V. Tereshatov, Valery Yu. Senichev V.V. Tereshatov, Vysokomol. soed., 37A, 1529 (1995). U.S. Aithal, T.M. Aminabhavi, R.H. Balundgi, and S.S. Shukla, JMS - Rev. Macromol. Chem. Phys., 30C (1), 43 (1990). P.J. Flory, Principles of Polymer Chemistry, Cornell Univ. Press, Ithaca, 1953. G.M. Bristow, and W.F. Watson, Trans. Faraday. Soc., 54, 1731 (1958). L.N. Mizerovsky, L.N. Vasnyatskaya, and G.M. Smurova, Vysokomol. Soed., 29A, 1512 (1987). D. Cohen, A. Siegman, and M. Narcis, Polym. Eng. Sci., 27, 286 (1987). E. Konton, G. Spathis, M. Niaounakis and V. Kefals, Colloid Polym. Sci., 26B, 636 (1990). V.V. Tereshatov, E.N. Tereshatova, and E.R. Volkova, Polym. Sci., 37A, 1157 (1995). E.E. Cluff, E.K. Gladding, and R. Pariser, J. Polym. Sci., 45, 341 (1960). Yu.Yu. Kercha, Z.V. Onishchenko, I.S. Kutyanina, and L.A. Shelkovnikova, Structural and Chemical Modification of Elastomers, Naukova Dumka, Kiev, 1989. Yu.Yu. Kercha, Physical Chemistry of Polyurethanes, Naukova Dumka, Kiev, 1979. V.N. Vatulev, S.V. Laptii, and Yu.Yu. Kercha, Infrared Spectra and Structure of Polyurethanes, Naukova Dumka, Kiev, 1987. S.V. Tereshatov, Yu.S. Klachkin, and E.N. Tereshatova, Plastmassy, 7, 43 (1998).

7

Solvent Transport Phenomena Many industrial processes rely on dissolution of raw materials and subsequent removal of solvents by various drying process. The formation of a solution and the subsequent solvent removal depends on a solvent transport phenomena which are determined by the properties of the solute and the properties of the solvent. Knowledge of the solvent movement within the solid matrix by a diffusion process is essential to design the technological processes. Many of the final properties, such as tribological properties, mechanical toughness, optical clarity, protection against corrosion, adhesion to substrates and reinforcing fillers, protective properties of clothing, the quality of the coated surface, toxic residues, morphology and residual stress, ingress of toxic substances, chemical resistance, depend not only on the material chosen but also on the regimes of technological processes. For these reasons, solvent transport phenomena are of interest to modern industry.

7.1 INTRODUCTION TO DIFFUSION, SWELLING, AND DRYING George Wypych ChemTec Laboratories, Inc., Toronto, Canada

Small molecule diffusion is the driving force behind movement of small molecules in and out of the solid matrix. Although both swelling and drying rely on diffusion, these processes are affected by the surfaces of solids, the concentration of small molecules in the surface layers, the morphology of the surface, and the interface between phases in which diffusion gradient exists. For these reasons, swelling and drying are treated as specific phenomena. 7.1.1 DIFFUSION The free-volume theory of diffusion was developed by Vrentas and Duda.1 This theory is based on the assumption that movement of a small molecule (e.g., solvent) is accompanied by a movement in the solid matrix to fill the free volume (hole) left by a displaced solvent molecule. Several important conditions must be described to model the process. These include the time scales of solvent movement and the movement of solid matrix (e.g. polymer segments, called jumping units), the size of holes which may fit both solvent molecules and jumping units, and the energy required for the diffusion to occur. The timescale of the diffusion process is determined by the use of the diffusion Deborah, number De, given by the following equation: De =

τM τD

[7.1.1]

340

George Wypych

where: τM τD

the molecular relaxation time the characteristic diffusion time

If the diffusion Deborah number is small (small molecular relaxation time or large diffusion time) molecular relaxation is much faster than diffusive transport (in fact, it is almost instantaneous).2 In this case the diffusion process is similar to simple liquids. For example, diluted solutions and polymer solutions above glass transition temperature fall in this category. If the Deborah number is large (large molecular relaxation time or small diffusion time), the diffusion process is described by Fickian kinetics and is denoted by an elastic diffusion process.1 The polymeric structure in this process is essentially unaffected and coefficients of mutual and self-diffusion become identical. Elastic diffusion is observed at low solvent concentrations below the glass transition temperature.2 The relationships below give the energy required for the diffusion process and compare the sizes of holes required for the solvent and polymer jumping unit to move within the system. The free-volume coefficient of self-diffusion is given by the equation:2

(

 γ ω V$ * + ω ξV$ * 1 1 2 2  E  D1 = Do exp − × exp −  $   RT  VFH 

)   

[7.1.2]

where: D1 Do E R T γ ω^ V* ξ $ V FH

self-diffusion coefficient pre-exponential factor energy per molecule required by the molecule to overcome attractive forces gas constant temperature overlap factor introduced to address the fact that the same free volume is available for more than one molecule mass fraction (index 1 for solvent, index 2 for polymer) specific free hole volume (indices the same as above) the ratio of the critical molar volume of the solvent jumping unit to the critical molar volume of the polymer jumping unit (see equation [7.1.3]) average hole free volume per gram of mixture.

ξ = V$1* / V$2* = V$1*M1 / V$2*M 2

[7.1.3]

where: M

molecular weight (1 - solvent, 2 - polymer jumping unit)

The first exponent in equation [7.1.2] is the energy term and the second exponent is the free-volume term. Figure 7.1.1 shows three regions of temperature dependence of free-volume: I - above glass transition temperature, II - close to transition temperature, and III - below the transition temperature. In the region I, the second term of the equation [7.1.2] is negligible and thus diffusion is energy-driven. In the region II both terms are significant. In the region III the diffusion is free volume-driven.3

7.1 Introduction to diffusion, swelling, and drying

341

0.7

8 10-7

Region II

III

-7

6 10 -1

5 10

2

4 10-7

-7

D, cm s

Diffusivity, arbitrary units

I 0.5 0.4 0.3

3 10-7

0.2

2 10

0.1

1 10

0

methanol acetone Me acetate benzene

7 10-7

0.6

-7 -7

0

0

0.2

0.4

0.6

0.8

1

Temperature, arbitrary units

7

8

9

10

11

12

3

10 /(K + T - T ) 22

g2

Figure 7.1.1. Temperature dependence of the solvent Figure 7.1.2. Free-volume correlation data for various self-diffusion coefficient. [Adapted, by permission, solvents. [Data from D Arnauld, R L Laurence, Ind. Eng. from D Arnauld, R L Laurence, Ind. Eng. Chem. Res., Chem. Res., 31(1), 218-28 (1992).] 31(1), 218-28 (1992).]

The mutual diffusion coefficient is given by the following equation: D=

D1 ω1 ω2 RT

 ∂µ 1    = D1Q  ∂ω1 T , p

[7.1.4]

where: D µ1 Q

mutual diffusion coefficient chemical potential of a solvent per mole thermodynamic factor.

These equations are at the core of diffusion theory and are commonly used to predict various types of solvent behavior in polymeric and other systems. One important reason for their wide application is that all essential parameters of the equations can be quantified and then used for calculations and modelling. The examples of data given below illustrate the effect of important parameters on the diffusion processes. Figure 7.1.2 shows the effect of temperature on the diffusivity of four solvents. The relationship between diffusivity and temperature is essentially linear. Only solvents having the smallest molecules (methanol and acetone) depart slightly from a linear relationship due to the contribution of the energy term. The diffusivity of the solvent decreases as temperature decreases. Several other solvents show a similar relationship.3 Figure 7.1.3 shows the relationship between the solvent’s molar volume and its activation energy. The activation energy increases as the solvent’s molar volume increases then levels off. The data show that the molar volume of a solvent is not the only parameter which affects activation energy. Flexibility and the geometry of solvent molecule also affect activation energy.3 Branched aliphatic solvents (e.g., 2-methyl-pentane, 2,3-dimethyl-butane) and substituted aromatic solvents (e.g., toluene, ethylbenzene, and xylene) show large departures from free volume theory predictions.

342

George Wypych

10-4

0.55 7

4

3

0.4

Diffusion coefficient, cm s

5

0.45

ξ

-1

6

0.5

8

2

0.35 0.3 0.25 0.2 1

0.15 30

40

50

60

70

80

90 100

Solvent molar volume at 0K, cm3 mole-1

10

-5

10

-6

10-7 mutual

10-8 10-9

self 0

20

40

60

80

100

PS, wt%

Figure 7.1.4. Mutual and self-diffusion coefficients Figure 7.1.3. Parameter ξ vs. solvent molar volume. 1 o methanol, 2 - acetone, 3 - methyl acetate, 4 - ethyl for polystyrene/toluene system at 110 C. [Adapted, by acetate, 5 - propyl acetate, 6 - benzene, 7 - toluene, 8 - permission, from F D Blum, S Pickup, R A Waggoner, ethylbenzene. [Adapted, by permission, from D Polym. Prep., 31 (1), 125-6 (1990).] Arnauld, R L Laurence, Ind. Eng. Chem. Res., 31(1), 218-28 (1992).]

1.4

20

Many experimental methods such as fluorescence, reflection Fourier transform 1.2 infrared, NMR, quartz resonators, and 15 acoustic wave admittance analysis, are 1 used to study diffusion of solvents.4-11 Spe10 0.8 cial models have been developed to study process kinetics based on experimental 0.6 5 data. Figure 7.1.4 shows the effect of con0.4 centration of polystyrene on mutual and 0 0.2 self-diffusion coefficients measured by acetone pulsed-gradient spin-echo NMR. The data 0 -5 0 20 40 60 80 100 show that the two coefficients approach each other at high concentrations of polyTime, min mer as predicted by theory.4 Figure 7.1.5. Flux vs. time for two solvents. [Adapted by Studies on solvent penetration permission from C F Fong, , Y Li, De Kee, J Bovenkamp, through rubber membranes (Figure 7.1.5), Rubber Chem. Technol., 71(2), 285-288 (1998).] show that in the beginning of the process, there is a lag time called break-through time. This is the time required for solvent to begin to penetrate the membrane. It depends on both the solvent and the membrane. Of solvents tested, acetone had the longest break-through time in natural rubber and toluene the longest in nitrile rubber. After penetration has started the flux of solvent increases rapidly and ultimately levels off.7 This study is relevant in testing the permeability of protective clothing and protective layers of coatings. 2

Flux, µg cm s

-1

benzene

343

1

1

0.8

0.8 Absorbance ratio

Absorbance ratio


0.6 0.4 43,000

0.2

0.6 0.4 286 µm 185 µm

0.2

8,000 0

0

50 100 150 200 250 300 350 400 1/2

t ,s

1/2

Figure 7.1.6. Absorbance ratio vs. exposure time to water for PMMA of different molecular weights. [Adapted, by permission, from I Linossier, F Gaillard, M Romand, J F Feller, J. Appl. Polym. Sci., 66, No.13, 2465-73 (1997).]

0.12

0 2 105

6 105 1 106 1/2 t /2L, s1/2 m -1

1.4 106

Figure 7.1.7. Absorbance ratio vs. exposure time to water for PMMA films of different thickness. [Adapted, by permission, from I Linossier, F Gaillard, M Romand, J F Feller, J. Appl. Polym. Sci., 66, No.13, 2465-73 (1997).]

-1

Diffusion rate, g g min

-1

Similar observations were made using internal reflection Fourier transform infra0.1 red to measure water diffusion in polymer films.9 Figure 7.1.6 shows that there is a 0.08 time lag between the beginning of immersion and water detection in polymer film. 0.06 This time lag increases as the molecular tp weight increases (Figure 7.1.6) and film 0.04 thickness increases (Figure 7.1.7). After an initial increase in water concentration, the 0.02 amount levels off. Typically, the effect of molecular weight on the diffusion of the 0 0 10 20 30 40 50 60 70 penetrant does not occur. High molecular weight polymer has a shift in the absorption Time, min peak from 1730 to 1723 cm-1 which is assoFigure 7.1.8. Relative diffusion rate vs. curing time. ciated with the hydrogen bonding of the [Adapted, by permission, from Jinhua Chen, Kang Chen, carbonyl group. Such a shift does not occur Hong Zhang, Wanzhi Wei, Lihua Nie, Shouzhuo Yao, J. in low molecular weight PMMA. Water can Appl. Polym. Sci., 66, No.3, 563-71 (1997).] move at a higher rate in low molecular weight PMMA. In some other polymers, this trend might be reversed if the lower molecular weight polymer has end groups which can hydrogen bond with water. Bulk acoustic wave admittance analysis was used to study solvent evaporation during curing.8 Three characteristic stages were identified: in the first stage viscosity increases accompanied by a rapid decrease in diffusion rate; in the second stage the film is formed, the

344

George Wypych

surface appears dry (this stage ends when the surface is dry) and the diffusion rate becomes very low; in the third stage solvent diffuses from the cured film. This is a slow process during which diffusion rate drops to zero. These changes are shown in Figure 7.1.8. The diffusion rate during drying decreases as the concentration of polymer (phenol resin) in varnish increases. Also, Light scattering the time to reach the slope change point in Fluorescence recovery diffusion/time relationship increases as the concentration of polymer increases.8 1 Two methods have been used to mea0.1 1 10 sure the diffusion coefficient of toluene in Polymer concentration, g cm-3 Figure 7.1.9. Self-diffusion coefficient vs. polystyrene mixtures of polystyrenes having two different molecular weights: one was dynamic concentration. [Adapted, by permission, from L Meistermann, M Duval, B Tinland, Polym. Bull., 39, light scattering and the other, fluorescence No.1, 101-8 (1997).] recovery after bleaching.10 The data show that the relationship between the diffusion coefficient and polymer concentration is not linear. The crossover point is shown in Figure 7.1.9. Below a certain concentration of polymer, the diffusion rate drops rapidly according to different kinetics. This is in agreement with the above theory (see Figure 7.1.1 and explanations for equation [7.1.2]). The slope exponent in this study was -1.5 which is very close to the slope exponent predicted by the theory of reptation (-1.75). The above data show that theoretical predictions are accurate when modelling diffusion phenomena in both simple and complicated mixtures containing solvents. 7.1.2 SWELLING Polymers differ from other solids because they may absorb large amounts of solvents without dissolving. They also undergo large deformations when relatively small forces are involved.12 Swelling occurs in a heterogeneous two phase system a solvent surrounding a swollen body also called gel. Both phases are separated by the phase boundary permeable to solvent.13 The swelling process (or solvent diffusion into to the solid) occurs as long as the chemical potential of solvent is large. Swelling stops when the potentials are the same and this point is called the swelling equilibrium. Swelling equilibrium was first recognized by Frenkel14 and the general theory of swelling was developed by Flory and Rehner.15,16 The general theory of swelling assumes that the free energy of mixing and the elastic free energy in a swollen network are additive. The chemical potential difference between gel and solvent is given by the equation: -8

2

Self-diffusion coefficient, 10 cm s

-1

10

(µ 1 − µ 10 ) = (µ 1 − µ 10 ) mix + (µ 1 − µ 10 ) el where: µ1 µ 10

chemical potential of gel chemical potential of solvent

[7.1.5]


345

The chemical potential is the sum of the terms of free energy of mixing and the elastic free energy. At swelling equilibrium, µ 1 = µ 10 , and thus the left hand term of the equation becomes zero. The equation [7.1.5] takes the following form: (µ 1 − µ 10 ) mix = −(µ 1 − µ 10 ) el = RT [ln(1 − v 2 ) − v 2 + χv 22 ]

[7.1.6]

where: v2 n1, n2 V1, V2 R T χ

= n2V2/(n1V1 + n2V2) volume fraction of polymer moles of solvent and polymer, respectively molar volumes of solvent and polymer, respectively gas constant absolute temperature Flory-Huggins, polymer-solvent interaction parameter.

The interaction between the solvent and solid matrix depends on the strength of such intermolecular bonds as polymer-polymer, solvent-solvent, and polymer-solvent. If interaction between these bonds is similar, the solvent will easily interact with polymer and a relatively small amount of energy will be needed to form a gel.12 The Hildebrand and Scatchard hypothesis assumes that interaction occurs between solvent and a segment of the chain which has a molar volume similar to that of solvent.12 Following this line of reasoning the solvent and polymer differ only in potential energy and this is responsible for their interaction and for the solubility of polymer in the solvent. If the potential energies of solvents and polymeric segments are similar they are readily miscible. In crosslinked polymers, it is assumed that the distance between crosslinks is proportional to the molecular volume of the polymer segments. This assumption is the basis for determining molecular mass between crosslinks from results of swelling studies. The result of swelling in a liquid solvent (water) is determined by equation:13 g/l

 ∂T     ∂w1  P

 ∂µ g  T 1   ∂w   1 T , P = ∆H1g / l

[7.1.7]

where: T w1 g l P µ 1g ∆H1g /l

thermodynamic (absolute) temperature mass fraction of solvent in gel at saturation concentration phase symbol (for gel) symbol for liquid pressure chemical potential of solvent in gel phase dependent on temperature = H1g − H-l01 is the difference between partial molar enthalpy of solvent (water) in gel and pure liquid solvent (water) in surrounding

Contrast this with the equation for water in the solid state (ice): g / cr

 ∂T     ∂w1  P where:

cr

 ∂µ g  T 1   ∂w   1 T , P = ∆H1g / cr phase symbol for crystalline solvent (ice)

[7.1.8]

346

George Wypych

A comparison of equations [7.1.7] and [7.1.8] shows that the slope and sign of swelling curve are determined by the quan300 tity ∆H1g / l or cr . Since the melting enthalpy of water is much larger than the transfer 290 enthalpy of water, the swelling curves of gel in liquid water are very steep. The sign of the slope is determined by the heat transPEO/water 280 PEO/ice fer of the solvent which may be negative, positive or zero depending on the quality of 270 solvent. The melting enthalpy is always positive and therefore the swelling curve in 260 the presence of crystalline solvent is flat 0.25 0.3 0.35 0.4 0.45 0.5 0.55 with a positive slope. A positive slope in Mass fraction of water temperatures below zero (for ice) means that gel has to deswell (release water to its Figure 7.1.10. Swelling of crosslinked polyurethane in water and ice. [Adapted, by permission, from B Hladik, surrounding, or dry out) as temperature S Frahn, W Borchard, Polym. Polym. Compos., 3, No.1, decreases.13 Figure 7.1.10 illustrates this. 21-8 (1995).] For practical purposes, simple equations are used to study swelling kinetics. The degree of swelling, α, is calculated from the following equation:17 Temperature, K

310

α=

V1 − V0 V0

[7.1.9]

where: V1 V0

volume of swollen solid at time t=t volume of unswollen solid at time t=0

The swelling constant, K, is defined by: K=

k1 α = k 2 1− α

[7.1.10]

where: k1 k2

rate constant of swelling process rate constant of deswelling process

This shows that the swelling process is reversible and in a dynamic equilibrium. The distance of diffusion is time-dependent: distance ∝ (time) n

[7.1.11]

The coefficient n is between 0.5 for Fickian diffusion and 1.0 for relaxation-controlled diffusion (diffusion of solvent is much faster than polymer segmental relaxation).18 This relationship is frequently taken literally19 to calculate diffusion distance from a measurement of the change of the linear dimensions of swollen material. The following equation is used to model changes based on swelling pressure measurements:

347

0.7

0.8

0.6

0.7

0.5

0.6

Degree of swelling

Degree of swelling


0.4 0.3 0.2 i-octane toluene

0.1 0

0

20

40

60

80

0.5 0.4 0.3 298K 303K 308K

0.2 0.1

100

0

0

20 40 60 80 100 120 Time, min

Time, min Figure 7.1.11. Swelling kinetics of EVA in toluene and i-octane. [Data from H J Mencer, Z Gomzi, Eur. Polym. J., 30, 1, 33-36, (1994).]

Π = Aϕ n = −

RT ln(1 − ϕ) + ϕ + χϕ 2 v1

[

Figure 7.1.12. Swelling kinetics of EVA in tetrahydrofuran at different temperatures. [Adapted, by permission, from H J Mencer, Z Gomzi, Eur. Polym. J., 30, 1, 33-36, (1994).]

]

[7.1.12]

where: Π A ϕ n v1 χ

osmotic pressure coefficient volume fraction of polymer in solution = 2.25 for good solvent and = 3 for Θ solvent molar volume of solvent Flory-Huggins, polymer-solvent interaction parameter

The above relationship is used in to study swelling by measuring shear modulus.20 Figure 7.1.11 shows swelling kinetic curves for two solvents. Toluene has a solubility parameter of 18.2 and i-octane of 15.6. The degree to which a polymer swells is determined by many factors, including the chemical structures of polymer and solvent, the molecular mass and chain flexibility of the polymer, packing density, the presence and density of crosslinks, temperature, and pressure. In the example presented in Figure 7.1.11 the solubility parameter has a strong influence on swelling kinetics.17 The effect of temperature on swelling kinetics is shown in Figure 7.1.12. Increasing temperature increases swelling rate. During the initial stages of the swelling process the rate of swelling grows very rapidly and then levels off towards the swelling equilibrium. In Figure 7.1.13, the diffusion distance is almost linear with time as predicted by equation [7.1.11]. The coefficient n of the equation was 0.91 meaning that the swelling process was relaxation rate controlled. Figure 7.1.14 shows the relationship between hydrogel swelling and pH. Hydrogels are particularly interesting because their swelling properties are controlled by the conditions around them (e.g. pH).21-23 Because they undergo controllable volume changes, they find applications in separation processes, drug delivery systems, immobilized enzyme sys-

348

George Wypych

40 Equilibrium degree of swelling

3.5

Diffusion distance, mm

3 2.5 2 1.5 1 0.5 0

0

2000 4000 6000 8000 10000 Time, min

35

30

25

2

3

4

5

6

7

8

pH Figure 7.1.13. Diffusion distance of 1,4-dioxane Figure 7.1.14. Effect of pH on the equilibrium swelling of into PVC vs. time. [Adapted, by permission, from hydrogel. [Data from M Sen, A Yakar, O Guven, Polymer, M Ercken, P Adriensens, D Vanderzande, J Gelan, 40, No.11, 2969-74 (1999).] Macromolecules, 28, 8541-8547, (1995).]

tems, etc. Maximum swelling is obtained at pH = 7. At this point there is complete dissociation of the acidic groups present in the hydrogel. The behavior of hydrogel can be modelled using Brannon-Peppas equation.22 Figure 7.1.15 shows the 1,1,2,2-tetrachoroethylene, TCE, uptake by amorphous poly(ether ether ketone), PEEK, as a function of time.24,25 The swelling behavior of PEEK in this solvent is very unusual - the sample mass is increased by 165% which is about 3 times more than with any other solvent. In addition, the solvent uptake by PEEK results in a change in optical properties of the solution from clear to opaque. A clear solution is typical of amorphous PEEK and the opaque solution of crystalline PEEK. It was previously suggested by Fokes and Tischler26 that polymethylmethacrylate forms weak complexes with various acid species in solution. This may also explain the unusual swelling caused by TCE. Because of the presence of C=O, C-O-C, and aromatic groups, PEEK acts as organic base. TCE is an electron acceptor due to electron-deficient atoms in the molecule.24 This interaction may explain the strong affinity of solvent and polymer. Figure 7.1.16 illustrates the effect of crystallization. Below 250oC, the carbonyl frequency decreases. But a more rapid decrease begins below 140oC which is glass transition temperature. Above 250oC, the carbonyl frequency increases rapidly. Above the glass transition temperature there is rapid crystallization process which continues until the polymer starts to melt at 250oC then it gradually reverts to its original amorphous structure. The presence of solvent aids in the crystallization process. 7.1.3 DRYING Solvent removal can be accomplished by one of three means: deswelling, drying or changes in the material’s solubility. The deswelling process, which involves the crystallization of solvent in the surrounding gel, was discussed in the previous section. Here attention is focused on drying process. The changes due to material solubility are discussed in Chapter 12.

349

1654

200

1652

-1

250

Carbonyl frequency, cm

Mass uptake, %


150 100 50 0

0

20

40

60

80

100 120

1650 1648 1646 1644

0

50 100 150 200 250 300 350 Temperature, oC

Time, min

Figure 7.1.15. The mass uptake of TCE by PEEK vs. Figure 7.1.16. The frequency of carbonyl stretching time. [Adapted, by permission, from B H Stuart, mode of PEEK vs. temperature. [Adapted, by D R Williams, Polymer, 35, No.6, 1326-8 (1994).] permission, from B H Stuart, D R Williams, Polymer, 36, No.22, 4209-13 (1995).]

100

Figure 7.1.1 can be discussed from a different perspective of results given in Fig80 ure 7.1.17. There are also three regions here: region 1 which has a low concentration of solid in which solvent evaporation is con60 trolled by the energy supplied to the system, region 2 in which both the energy supplied 40 to the system and the ability of polymer to take up the free volume vacated by solvent are important, and region 3 where the pro20 cess is free volume controlled. Regions 2 and 3 are divided by the glass transition 0 0 0.2 0.4 0.6 0.8 1 temperature. Drying processes in region 3 and to some extent in region 2 determine the Time, arbitrary units physical properties of dried material and the amount of residual solvent remaining in the Figure 7.1.17. Solvent concentration vs. drying time. product. A sharp transition between region 2 and 3 (at glass transition temperature) might indicate that drying process is totally homogeneous but it is not and this oversimplifies the real conditions at the end of drying process. The most realistic course of events occurring close to the dryness point is presented by these four stages:27,28 • elimination of the volatile molecules not immobilized by the adsorption onto the polymer • elimination of adsorbed molecules from the polymer in its rubbery state Solvent concentration, %

solution

rubbery

glassy

350

George Wypych

• evaporation-induced self association of the polymer with progressive entrapment of adsorbed volatile molecules in the glassy microdomains (during transition from a rubbery to a glassy state) • elimination of residual molecules entrapped in the polymer. The last two stages are discussed in Chapter 16 which deals with residual solvent. This discussion concentrates on the effect of components on the drying process and the effect of the drying process on the properties of the product. A schematic of the drying process is represented in Figure 7.1.18. The material to be dried is placed on an impermeable substrate. The material consists of solvent and semicrystalline polymer which contains a certain initial fraction of amorphous and crystalline domains. The presence of crystalline domains complicates the proFigure 7.1.18. Schematic representation of drying a polymer slab. [Adapted, by permission, from M O Ngui, cess of drying because of the reduction in S K Mallapragada, J. Polym. Sci.: Polym. Phys. Ed., 36, diffusion rate of the solvent. Evaporation No.15, 2771-80 (1998).] of solvent causes an inward movement of material at the surface and the drying process may change the relative proportions of amorphous and crystalline domains.29 Equations for the change in thickness of the material and kinetic equations which relate composition of amorphous and crystalline domains to solvent concentration are needed to quantify the rate of drying. The thickness change of the material during drying is given by the equation: v1

dL  ∂v 1  = D  dt  ∂x  x = L

[7.1.13]

where: v1 L t D x

volume fraction of solvent thickness of slab as in Figure 7.1.18 time diffusion coefficient coordinate of thickness

The rate of change of crystalline volume fraction is given by the equation: ∂v 2 c = k1v 1 ∂t

[7.1.14]

where: v2c k1

volume fraction of crystalline phase rate change of crystalline phase proportional to folding rate

The rate of change of amorphous volume fraction is given by the equation: ∂v 2 a ∂  ∂v 1  = D  − k1v 1 ∂t ∂x  ∂x 

[7.1.15]

351

1

50

0.8

40

Water volume fraction, %

Residual water fraction


0.6 0.4 0.2 0

0

20 10

0

50 100 150 200 250 300 350 400 Time, min

Time, min Figure 7.1.19. Fraction of water remaining in PVA as a function of drying time at 23oC. [Data from M O Ngui, S K Mallapragada, J. Polym. Sci.: Polym. Phys. Ed., 36, No.15, 2771-80 (1998).]

30

0

50 100 150 200 250 300 350

41.8 44.9

Figure 7.1.20. Water volume fraction vs. drying time for PVA of different crystallinity. [Adapted, by permission, from M O Ngui, S K Mallapragada, J. Polym. Sci. Polym. Phys. Ed., 36, No.15, 2771-80 (1998).]

where: v2a

volume fraction of amorphous phase

The rate of drying process is determined by the diffusion coefficient: D = D0 [exp(α D v 1 )](1 − v 2 c ) / τ

[7.1.16]

where: D0 αD τ

initial diffusion coefficient dependent on temperature constant which can be determined experimentally for spin echo NMR studies30 constant equal to 1 for almost all amorphous polymers (v2c ≤ 0.05) and 3 for semi-crystalline polymers

According to this equation, the coefficient of diffusion decreases as crystallinity increases because the last term decreases and τ increases. Figure 7.1.19 shows that the fraction of solvent (water) decreases gradually as drying proceeds. Once the material reaches a glassy state, the rate of drying rapidly decreases. This is the reason for two different regimes of drying. Increasing temperature increases the rate of drying process.29 Figure 7.1.20 shows that even a small change in the crystallinity of the polymer significantly affects drying rate. An increase in molecular weight has similar effect (Figure 7.1.21). This effect is due both to lower mobility of entangled chains of the higher molecular weight and to the fact that higher molecular weight polymers are more crystalline. Figure 7.1.22 shows that drying time increases as the degree of polymer crystallinity increases. Increased crystallinity slows down the diffusion rate and thus the drying process. The physical properties of liquids, such as viscosity (Figure 7.1.23) and surface tension (Figure 7.1.24), also change during the evaporation process. The viscosity change for this system was a linear function of the amount of solvent evaporated.31 This study on waterborne coatings showed that the use of a cosolvent (e.g. i-butanol) caused a reduction in

352

George Wypych

49

64,000 17,600

40

Degree of crystallinity, %

Water volume fraction, %

50

30 20 10 0

0

50

47

45

43

41

100 150 200 250 300

0

volume fraction vs. drying time molecular weight. [Adapted, by O Ngui, S K Mallapragada, J. Phys. Ed., 36, No.15, 2771-80

400

48

Surface tension, mN m

-1

1.2 Viscosity, Pa s

300

Figure 7.1.22. Degree of crystallinity vs. PVA drying time at 25oC. [Adapted, by permission, from M O Ngui, S K Mallapragada, J. Polym. Sci.: Polym. Phys. Ed., 36, No.15, 2771-80 (1998).]

1.4

1 0.8 0.6 0.4

200 Time, min

Time, min Figure 7.1.21. Water for PVA of different permission, from M Polym. Sci.: Polym. (1998).]

100

0

0.02 0.04 0.06 0.08 0.1

0.12

Fraction of total evaporation Figure 7.1.23. Viscosity vs. fraction of total evaporation of water from waterborne coating. [Data from S Kojima, T Moriga, Polym. Eng. Sci., 35, No.13, 1098-105 (1995).]

46 44 42 40 H2O

38 36

BuOH 0

0.04

0.08

0.12

Fraction of total evaporation Figure 7.1.24. Surface tension vs. fraction of total evaporation of solvent from waterborne coating. [Data from S Kojima, T Moriga, Polym. Eng. Sci., 35, No.13, 1098-105 (1995).]

overall viscosity. Micelles of smaller size formed in the presence of the cosolvent explain lower viscosity.31 Figure 7.1.24 shows that the surface tension of system containing a cosolvent (i-butanol) increases as the solvents evaporate whereas the surface tension of a system containing only water decreases. The increase of surface tension in the system containing cosolvent is due to the preferential evaporation of the cosolvent from the mixture.31


0.12

Table 7.1.1. Experimentally determined initial evaporation rates of waterborne coating containing a variety of solvents at 5% level (evaporation at 25oC and 50% RH)

HO

-1

2

Initial evaporation rate, % s

353

H2O+BuOH

0.1 0.08

Co-solvent

0.06 0.04 0.02 0

10

20 30 40 50 60 Relative humidity, %

Initial evaporation rate, µg cm-2 s-1

none

3.33

methyl alcohol

4.44

ethyl alcohol

3.56

n-propyl alcohol

4.00

70 n-butyl alcohol

3.67

i-butyl alcohol

3.67

n-amyl alcohol

3.33

n-hexyl alcohol

3.22

ethylene glycol mono-butyl ester

3.22

Figure 7.1.25. Initial evaporation rate from waterborne coating vs. relative humidity for two solvent systems. [Data from S Kojima, T Moriga, Polym. Eng. Sci., 35, No.13, 1098-105 (1995).]

Table 7.1.1 shows the effect of cosolvent ethylene glycol 3.33 addition on the evaporation rate of solvent mono-hexyl ester mixture. butyl carbinol 3.11 The initial rate of evaporation of 4.89 solvent depends on both relative humidity methyl-i-butyl ketone and cosolvent presence (Figure 7.1.25). As relative humidity increases the initial evaporation rate decreases. The addition of cosolvent doubles the initial evaporation rate. In convection drying, the rate of solvent evaporation depends on airflow, solvent partial pressure, and temperature. By increasing airflow or temperature, higher process rates can be achieved but the risk of skin and bubble formation is increased. As discussed above, Vrentas-Duda free-volume theory is the basis for predicting solvent diffusion, using a small number of experimental data to select process conditions. The design of a process and a dryer which uses a combination of convection heat and radiant energy is a more complex process. Absorption of radiant energy is estimated from the Beer’s Law, which, other than for the layers close to the substrate, predicts:32 Q r ( ξ) = I 0 α exp[−α(βh − ξ)] where: Qr ξ I0 α β h

radiant energy absorption distance from substrate intensity of incident radiation volumetric absorption coefficient fractional thickness of the absorbing layer next to the substrate thickness

[7.1.17]

354

George Wypych

1 Evaporated solvent fraction

A B

40

2

Diffusion coefficient, cm s

-1

50

30 20 10 0 30

40

50

60

70 o

Temperature, C

80

90

0.8 0.6 0.4 2 mm 1 mm

0.2 0

0

2

4

6

8

10

12

14

Time, h

Figure 7.1.26. Diffusion coefficients for two Figure 7.1.27. Effect of film thickness on evaporation polyurethanes (A & B) dissolved in N,N-dimethyl- rate. [Adapted, by permission from G A Abraham, T R acetamide vs. temperature. [Data from G A Abraham, T Cuadrado, International Polym. Process., 13, No.4, R Cuadrado, International Polym. Process., 13, No.4, Dec.1998, p.369-78.] Dec.1998, 369-78.]

The radiant energy delivered to the material depends on the material’s ability to absorb energy which may change as solvent evaporates. Radiant energy • compensates for energy lost due to the evaporative cooling - this is most beneficial during the early stages of the process • improves performance of the dryer when changes to either airflow rate or energy supplied are too costly to make. Radiant energy can be used to improve process control. For example, in a multilayer coating (especially wet on wet), radiant energy can be used to regulate heat flow to each layer using the differences in their radiant energy absorption and coefficients of thermal conductivity and convective heat transfer. Experimental work by Cairncross et al.32 shows how a combination dryer can be designed and regulated to increase the drying rate and eliminate bubble formation (more information on the conditions of bubble formation is included in Section 7.2 of this chapter). Shepard33 shows how drying and curing rates in multilayer coating can be measured by dielectric analysis. Koenders et al.34 gives information on the prediction and practice of evaporation of solvent mixtures. Vrentas and Vrentas papers provide relevant modelling studies.35,36 These studies were initiated to explain the earlier observations by Crank37 which indicated that maintaining a slightly increased concentration of solvent in the air flowing over a drying material may actually increase the evaporation rate. Modelling of the process shows that although the diffusion of solvent cannot be increased by an increased concentration of solvent on the material’s surface, an increased concentration of solvent in the air may be beneficial for the evaporation process because it prevents the formation of skin which slows down solvent diffusion. Analysis of solvent evaporation from paint and subsequent shrinkage38 and drying of small particles obtained by aerosolization39 give further insight into industrial drying pro-


355

cesses. The two basic parameters affecting drying rate are temperature and film thickness. Figure 7.1.26 shows that the diffusion coefficient increases as temperature increases. Figure 7.1.27 shows that by reducing film thickness, drying time can be considerably reduced.40 Further information on the design and modelling of drying processes can be found in a review paper41 which analyzes drying process in multi-component systems.

REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

J S Vrentas and J.L. Duda, J. Polym. Sci., Polym. Phys. Ed., 15, 403 (1977); 17, 1085, (1979). J S Vrentas, C M Vrentas, J. Polym. Sci., Part B: Polym Phys., 30(9), 1005-11 (1992). D Arnauld, R L Laurence, Ind. Eng. Chem. Res., 31(1), 218-28 (1992). F D Blum, S Pickup, R A Waggoner, Polym. Prep., 31 (1), 125-6 (1990). C Bouchard, B Guerrier, C Allain, A Laschitsch, A C Saby, D Johannsmann, J. Appl. Polym. Sci., 69, No.11, 2235-46 (1998). G Fleischer, J Karger, F Rittig, P Hoerner, G Riess, K Schmutzler, M Appel, Polym. Adv. Technol., 9, Nos.10-11, 700-8 (1998). C F Fong, , Y Li, De Kee, J Bovenkamp, Rubber Chem. Technol., 71(2), 285-288 (1998). Jinhua Chen, Kang Chen, Hong Zhang, Wanzhi Wei, Lihua Nie, Shouzhuo Yao, J. Appl. Polym. Sci., 66, No.3, 563-71 (1997). I Linossier, F Gaillard, M Romand, J F Feller, J. Appl. Polym. Sci., 66, No.13, 2465-73 (1997). L Meistermann, M Duval, B Tinland, Polym. Bull., 39, No.1, 101-8 (1997). J S Qi, C Krishnan, J A Incavo, V Jain, W L Rueter, Ind. Eng. Chem. Res., 35, No.10, 3422-30 (1996). Z Hrnjak-Murgic, J Jelencic, M Bravar, Angew. Makromol. Chem., 242, 85-96 (1996). B Hladik, S Frahn, W Borchard, Polym. Polym. Compos., 3, No.1, 21-8 (1995). J Frenkel, Rubber Chem. Technol., 13, 264 (1940). P J Flory, J Rehner, J. Chem. Phys., 11, 521 (1943). P J Flory, J. Chem. Phys., 18, 108 (1950). H J Mencer, Z Gomzi, Eur. Polym. J., 30, 1, 33-36, (1994). A G Webb, L D Hall, Polymer, 32(16), 2926-38 (1991). M Ercken, P Adriensens, D Vanderzande, J Gelan, Macromolecules, 28, 8541-8547, (1995). F Horkay, A M Hecht, E Geissler, Macromolecules, 31, No.25, 8851-6 (1998). M Sen, O Guven, Polymer, 39, No.5, 1165-72 (1998). M Sen, A Yakar, O Guven, Polymer, 40, No.11, 2969-74 (1999). Y D Zaroslov, O E Philippova, A R Khokhlov, Macromolecules, 32, No.5, 1508-13 (1999). B H Stuart, D R Williams, Polymer, 35, No.6, 1326-8 (1994). B H Stuart, D R Williams, Polymer, 36, No.22, 4209-13 (1995). F M Fowkes, D O Tischler, J. Polym. Sci., Polym. Chem. Ed., 22, 547 (1984). L A Errede, Macromol. Symp., 114, 73-84 (1997). L A Errede, P J Henrich, J N Schroepfer, J. Appl. Polym. Sci., 54, No.5, 649-67 (1994). M O Ngui, S K Mallapragada, J. Polym. Sci.: Polym. Phys. Ed., 36, No.15, 2771-80 (1998). N A Peppas, J C Wu, E D von Meerwall, Macromolecules, 27, 5626 (1994). S Kojima, T Moriga, Polym. Eng. Sci., 35, No.13, 1098-105 (1995). R A Cairncross, S Jeyadev, R F Dunham, K Evans, L F Francis, L Scriven, J. Appl. Polym. Sci., 58, No.8, 1279-90 (1995). D C Shepard, J. Coat. Technol., 68, No.857, 99-102 (1996). L Khal, E Juergens, J Petzoldt, M Sonntag, Urethanes Technology, 14, No.3, 23-6 (1997). J S Vrentas, C M Vrentas, J. Polym. Sci., Part B: Polym Phys., 30(9), 1005-11 (1992). J S Vrentas, C M Vrentas, J. Appl. Polym. Sci., 60, No.7, 1049-55 (1996). J Crank, Proc. Phys. Soc., 63, 484 (1950). L Ion, J M Vergnaud, Polym. Testing, 14, No.5, 479-87 (1995). S Norasetthekul, A M Gadalla, H J Ploehn, J. Appl. Polym. Sci., 58, No.11, 2101-10 (1995). G A Abraham, T R Cuadrado, International Polym. Process., 13, No.4, Dec.1998, 369-78. Z Pakowski, Adv. Drying, 5, 145-202 (1992).

356

Semyon Levitsky, Zinoviy Shulman

7.2 BUBBLES DYNAMICS AND BOILING OF POLYMERIC SOLUTIONS Semyon Levitsky Negev Academic College of Engineering, Israel

Zinoviy Shulman A.V. Luikov Heat and Mass Transfer Institute, Belarus

7.2.1 RHEOLOGY OF POLYMERIC SOLUTIONS AND BUBBLE DYNAMICS 7.2.1.1 Rheological characterization of solutions of polymers Solutions of polymers exhibit a number of unusual effects in flows.1 Complex mechanical behavior of such liquids is governed by qualitatively different response of the medium to applied forces than low-molecular fluids. In hydrodynamics of polymers this response is described by rheological equation that relates the stress tensor, σ, to the velocity field. The latter is described by the rate-of-strain tensor, e ∂v j 1  ∂v e ij =  i +  2  ∂x j ∂x i where: vi xi

   

[7.2.1]

r components of the velocity vector, v Cartesian coordinates (i, j = 1, 2, 3)

The tensors σ and e include isotropic, p, ekk and deviatoric, τ, s contributions: σ = −pI + τ,

r 1 1 p = − σ kk , e = e kk I + s, e kk = ∇ ⋅ v 3 3

[7.2.2]

where: I ∇

unit tensor Hamiltonian operator

For incompressible fluid ∇ ⋅ v = 0, e = s and rheological equation can be formulated in the form of the τ- dependence from e. Compressibility of the liquid must be accounted for in fast dynamic processes such as acoustic waves propagation, etc. For compressible medium the dependence of pressure, p, on the density, ρ, and the temperature, T, should be specified by equation, p = p(ρ,T), that is usually called the equation of state. The simplest rheological equation corresponds to incompressible viscous Newtonian liquid and has the form τ = 2η 0 e

[7.2.3]

where: η0

viscosity coefficient

Generalizations of the Newton’s flow law [7.2.3] for polymeric liquids are aimed to describe in more or less details the features of their rheological behavior. The most important among these features is the ability to accumulate elastic deformation during flow and thus to exhibit the memory effects. At first we restrict ourselves to the case of small deformation rates to discuss the basic principles of the general linear theory of viscoelasticity

7.2 Bubbles dynamics and boiling

357

based upon thermodynamics of materials with memory.2 The central idea of the theory is the postulate that instantaneous stresses in a medium depend on the deformation history. This suggestion leads to integral relationship between the stress and rate-of-strain tensors t

τ ij = 2 ∫ G1(t − t ′)s ij (t ′)dt ′, −∞

t

t

−∞

−∞

σ kk = −3p 0 + 3 ∫ G 2 (t − t ′)ekk (t ′)dt ′ − 3 ∫ G 3 (t − t ′)

∂θ dt ′ [7.2.4] ∂t ′

where: = T - T0 deviation of the temperature from its equilibrium value θ p0 equilibrium pressure G1,G2,G3 relaxation functions

The relaxation functions G1(t) and G2(t) satisfy restrictions, following from the entropy production inequality ∂G1 ≤ 0, ∂t

∂G 2 ≤0 ∂t

[7.2.5]

Additional restriction on G1,2(t) is imposed by the decaying memory principle3 that has clear physical meaning: the state of a medium at the present moment of time is more dependent on the stresses arising at t = t2 than on that stresses arising at t = t1 if t1 < t2. This principle implies that the inequality ∂ 2G1,2 / ∂t 2 ≥ 0 must be satisfied. The necessary condition for the viscoelastic material to be a liquid means

(

)

lim G1 (t) = 0

[7.2.6]

t→∞

Unlike G1(t), the functions G2(t) and G3(t) contain non-zero equilibrium components, such as lim G 2 (t) = G 20 , t→∞

lim G 3 (t) = G 30 t→∞

[7.2.7]

From the physical point of view this difference between G1 and G2,3 owes to the fact that liquid possesses a finite equilibrium bulk elasticity. Spectral representations of relaxation functions, accounting for [7.2.5] - [7.2.7], have form (G10 = 0) ∞

G i (t) = G i 0 + ∫ Fi (λ) exp(−t / λ)dλ

[7.2.8]

0

where: λ Fi(λ)

relaxation time spectrum of relaxation times (i = 1, 2, 3)

Equation [7.2.8] defines the functions Gi(t) for continuous distribution of relaxation times. For a discrete spectrum, containing ni relaxation times, the distribution function takes the form ni

Fi (λ) = ∑ G ik δ(λ − λ ik ) k =1

[7.2.9]

358


where: partial modules, corresponding to λik Dirac delta function

Gik δ(λ - λik)

In this case the integration over the spectrum in Equation [7.2.8] is replaced by summation over all relaxation times, λik. For polymeric solutions, as distinct to melts, it is convenient to introduce in the right-hand sides of equations [7.2.4] additional terms, 2ηssij and 3ρ 0 ηvekk, which represent contributions of shear, ηs, and bulk, ηv, viscosities of the solvent. The result has the form t ∞  t − t′  τ ij = 2 ∫ ∫ F1 (λ) exp −  s ij (t ′)dλdt ′ + 2ηs s ij λ   −∞ 0 t

p = p 0 − G20 ∫ e kk (t ′)dt ′ − −∞

t ∞

[7.2.10]

 t − t ′  e kk (t ′)dλdt ′ + G30 θ + λ 

∫ ∫ F (λ) exp − 2

−∞ 0

t ∞  t − t ′  ∂θ(t ′) + ∫ ∫ F3 (λ) exp − dλdt ′ − ρ 0 ηv e kk ,  λ  ∂t ′  −∞ 0

p = −1 / 3σ kk

[7.2.11]

Equilibrium values of bulk and shear viscosity, ηb and ηp, can be expressed in terms of the relaxation spectra, F1 and F2, as:1 ∞

∞

0

0

η p − ηs = ∫ λF1 (λ)dλ, η b − ηv = ∫ λF2 (λ)dλ

[7.2.12]

In the special case when relaxation spectrum, F1(λ), contains only one relaxation time, λ11, equation [7.2.10] yields t  t −t′   s ij (t ′)dt ′ + 2ηs s ij , G11 = (η p − ηs ) / λ11 τ ij = 2G11 ∫ exp −  λ11  −∞

[7.2.13]

At ηs = 0 the integral equation [7.2.13] is equivalent to the linear differential Maxwell equation τ ij + λ11 &τ ij = 2η p s ij

[7.2.14]

Setting ηs = λ2η0/λ1, λ1 = λ11, ηp = η0, where λ2 is the retardation time, one can rearrange equation [7.2.14] to receive the linear Oldroyd equation3

(

)

τ ij + λ1 &τ ij = 2η 0 s ij + λ ij s& ij ,

λ1 ≥ λ 2

[7.2.15]

Thus, the Oldroyd model represents a special case of the general hereditary model [7.2.10] with appropriate choice of parameters. Usually the maximum relaxation time in the spectrum is taken for λ1 in equation [7.2.15] and therefore it can be used for quantitative description and estimates of relaxation effects in non-steady flows of polymeric systems. Equation [7.2.11], written for quasi-equilibrium process, helps to clarify the meaning of the modules G20 and G30. In this case it gives


359

p = p 0 + G 20 ρ −01 (ρ − ρ 0 ) + G 30 θ

[7.2.16]

Thermodynamic equation of state for non-relaxing liquid at small deviations from equilibrium can be written as follows  ∂p  ∂p p = p 0 +   (ρ − ρ 0 ) +   θ  ∂T  ρ = ρ 0  ∂ρ T =T 0

[7.2.17]

The thermal expansion coefficient, α, and the isothermal bulk modulus, Kis, are defined as4  ∂p   ∂ρ  α = −ρ 0  , K is = ρ 0     ∂T  ρ = ρ 0  ∂ρ T =T 0

[7.2.18]

Therefore, equation [7.2.17] can be rewritten in the form p = p 0 + K is ρ −01 (ρ − ρ 0 ) + αK is θ

[7.2.19]

From [7.2.16] and [7.2.18] it follows that G20 = Kis, G30 = αKis. In rheology of polymers complex dynamic modulus, Gk*, is of special importance. It is introduced to describe periodic deformations with frequency, ω, and defined according to: ∞

G *k = ∫

Fk (λ)(ωλ)(ωλ + i )dλ 1 + (ωλ)

0

[7.2.20]

2

Equations of motion of the liquid follow from momentum and mass conservation laws. In the absence of volume forces they mean: r dv [7.2.21] ρ =∇ ⋅ σ dt r dρ + ρ∇ ⋅ v = 0 dt

[7.2.22]

For polymeric solution the stress tensor, σ, is defined according to [7.2.10], [7.2.11]. To close the system, it is necessary to add the energy conservation law to equations [7.2.21], [7.2.22]. In the case of liquid with memory it has the form2 k∇ 2 θ − T0

∂ ∂t

t

∫ G (t − t ′)e (t ′)dt ′ = T 3

−∞

kk

t

0

∂ ∂θ G 4 (t − t ′) dt ′ ∫ ∂t −∞ ∂t ′

∞  t − t′  −1 G 4 (t − t ′) = G 40 − ∫ F4 (λ) exp −  d λ, G 40 = ρ 0 c vT0 λ   0

[7.2.23]

360


where: k cv

heat conductivity of the liquid equilibrium isochoric specific heat capacity

Equations [7.2.10], [7.2.11], [7.2.21] - [7.2.23] constitute complete set of equations for linear thermohydrodynamics of polymeric solutions. The non-linear generalization of equation [7.2.4] is not single. According to the Kohlemann-Noll theory of a “simple” liquid,5 a general nonlinear rheological equation for a compressible material with memory may be represented in terms of the tensor functional that defines the relationship between the stress tensor, σ, and the deformation history. The form of this functional determines specific non-linear rheological model for a hereditary medium. Such models are numerous, the basic part of encountered ones can be found elsewhere.1,6-8 It is important, nevertheless, that in most cases the integral rheological relationships of such a type for an incompressible liquid may be brought to the set of the first-order differential equations.9 Important special case of this model represents the generalized Maxwell’s model that includes the most general time-derivative of the symmetrical tensor τ = ∑ τ (k ), k

Fabcτ =

τ ( k ) + λ k Fabc τ ( k ) = 2η k e

Dτ + a(τ ⋅ e + e ⋅ τ ) + bItr (τ ⋅ e) + ce tr (τ ), Dt

[7.2.24] Dτ dτ = − w ⋅ τ + τ ⋅ w, Dt dt

r r w = 1 ∇v − ∇v T 2

(

)

where: D/Dt d/dt tr w r ∇v T λk, ηk

Jaumann’s derivative1 ordinary total derivative trace of the tensor, trτ = τkk vorticity tensor r transpose of the tensor ∇v parameters, corresponding to the Maxwell-type element with the number k

At a = -1, b = c = 0, equations [7.2.24] correspond to the Maxwell liquid with a discrete spectrum of relaxation times and the upper convective time derivative.3 For solution of polymer in a pure viscous liquid, it is convenient to represent this model in such a form that the solvent contribution into total stress tensor will be explicit: τ = ∑ τ ( k ) + 2ηs e, k

Dτ ( k )  τ (k ) + λ k  − τ ( k ) ⋅ e + e ⋅ τ ( k )  = 2η k e  Dt 

(

)

[7.2.25]

To select a particular nonlinear rheological model for hydrodynamic description of the fluid flow, it is necessary to account for kinematic type of the latter.7 For example, the radial flows arising from the bubble growth, collapse or pulsations in liquid belong to the elongational type.3 Therefore, the agreement between the experimental and theoretically predicted dependencies of elongational viscosity on the elongational deformation rate should be a basic guideline in choosing the model. According to data10-13 the features of elongational viscosity in a number of cases can be described by equations [7.2.25]. More simple version of equation [7.2.25] includes single relaxation time and additional parameter 1/2 ≤ α ≤ 1, controlling the input of nonlinear terms:7


τ = τ (1) + τ ( 2 ) ,

361

  Dτ (1) τ (1) + λ  − α τ (1) ⋅ e + e ⋅ τ (1)  = 2ηβe,   Dt

(

)

τ ( 2 ) = 2η(1− β)e [7.2.26]

Parameter β governs the contribution of the Maxwell element to effective viscosity, η,(Newtonian viscosity of the solution). Equation [7.2.26] is similar to the Oldroyd-type equation [7.2.15] with the only difference that in the former the upper convective derivative is used to account for nonlinear effects instead of partial derivative, ∂ / ∂t. Phenomenological parameters appearing in theoretical models can be found from appropriate rheological experiments.6 Certain parameters, the most important being relaxation times and viscosities, can be estimated from molecular theories. According to molecular theory, each relaxation time λk is relative to mobility of some structural elements of a polymer. Therefore, the system as a whole is characterized by the spectrum of relaxation times. Relaxation phenomena, observed at a macroscopic level, owe their origin to the fact that response of macromolecules and macromolecular blocks to different-in-rate external actions is described by different parts of their relaxation spectrum. This response is significantly affected by temperature - its increase “triggers” the motion of more and more complex elements of the macromolecular hierarchy (groups of atoms, free segments, coupled segments, etc). The most studied relaxation processes from the point of view of molecular theories are those governing relaxation function, G1(t), in equation [7.2.4]. According to the Rouse theory,1 a macromolecule is modeled by a bead-spring chain. The beads are the centers of hydrodynamic interaction of a molecule with a solvent while the springs model elastic linkage between the beads. The polymer macromolecule is subdivided into a number of equal segments (submolecules or subchains) within which the equilibrium is supposed to be achieved; thus the model does not permit to describe small-scale motions that are smaller in size than the statistical segment. Maximal relaxation time in a spectrum is expressed in terms of macroscopic parameters of the system, which can be easily measured: λ11 =

6 (η p − ηs ) M π 2 cRGT

[7.2.27]

where: M c RG

molecular mass of the polymer concentration of polymer in solution universal gas constant

The other relaxation times are defined as λ1k = λ11/k2. In Rouse theory all the modules G1k are assumed to be the same and equal to cRGT/M. In the Kirkwood-Riseman-Zimm (KRZ) model, unlike Rouse theory, the hydrodynamic interaction between the segments of a macromolecular chain is accounted for. In the limiting case of a tight macromolecular globe, the KRZ theory gives the expression for λ11 that is similar to [7.2.27]: λ11 =

0.422 (η p − ηs ) M cRGT

[7.2.28]

362


The differences between other relaxation times in the both spectra are more essential: the distribution, predicted by the KRZ model, is much narrower than that predicted by the Rouse theory. The KSR and Rouse models were subjected to numerous experimental tests. A reasonably good agreement between the theoretical predictions and experimental data was demonstrated for a variety of dilute polymeric solutions.14 Further advance in the molecular-kinetic approach to description of relaxation processes in polymeric systems have brought about more sophisticated models.15,16 They improve the classical results by taking into account additional factors and/or considering diverse frequency, temperature, and concentration ranges, etc. For the aims of computer simulation of the polymeric liquid dynamics in hydrodynamic problems, either simple approximations of the spectrum, F1(λ), or the model of subchains are usually used. Spriggs law17 is the most used approximation λ1k = λ11 / k Z , z ≥ 2

[7.2.29]

The molecular theory predicts strong temperature dependence of the relaxation characteristics of polymeric systems that is described by the time-temperature superposition (TTS) principle.18 This principle is based on numerous experimental data and states that with the change in temperature the relaxation spectrum as a whole shifts in a self-similar manner along t axis. Therefore, dynamic functions corresponding to different temperatures are similar to each other in shape but are shifted along the frequency axis by the value aT; the latter is named the temperature-shift factor. With ωaT for an argument it becomes possible to plot temperature-invariant curves Re{G1*(ω aT)} and Im{G1*(ω aT)}. The temperature dependence of aT is defined by the formula aT =

ρ(T0 )T0 (η p (T ) − ηs (T ))

[7.2.30]

ρ(T )T (η p (T0 ) − ηs (T0 ))

The dependence of viscosity on the temperature can be described by the activation theory19

[

]

ηp = ηp 0 exp E p (RGT 0 ) (T 0 / T − 1) , −1

[

]

ηs = ηs 0 exp E s (RGT 0 ) (T 0 / T − 1) −1

[7.2.31]

where: Ep, Es

activation energies for the solution and solvent, respectively

The Es value is usually about 10 to 20 kJ/mol. For low-concentrated solutions of polymers with moderate molecular masses, the difference between these two activation energies, ∆E = Ep - Es, does not exceed usually 10 kJ/mol.18,20 For low-concentrated solutions of certain polymers in thermodynamically bad solvents negative ∆E values were reported.20 The Newtonian viscosity of solution related to the polymer concentration can be evaluated, for example, using Martin equation18 η p / ηs = 1 + c~ exp(k M c~), c~ = c[η] where:

~c [η]

reduced concentration of polymer in the solution intrinsic viscosity

[7.2.32]


363

Martin equation is usually valid in the range of reduced concentrations, ~ c ≤ 10. For evaluation of [η], the Mark-Houwink relationship21 is recommended

[η] = KM a

[7.2.33]

where K and a are constants for a given polymer-solvent pair at a given temperature over a certain range of the molecular mass variation. The parameter a (the Mark-Houwink exponent) lies in the range 0.5 to 0.6 for solutions of flexible chains polymers in thermodynamically bad solvents and in the range 0.7- 0.8 for good solvents. For the former ones the constant K ≈ 10-2 (if the intrinsic viscosity [η] is measured in cm3/g), while for the latter K≈10-3. Thus, the spectral functions, F1(λ), are comprehensively studied both experimentally and theoretically. The behavior of relaxation functions, G2(t) and G3(t), is still much less known. The properties of the function, G2(t), were mainly studied in experiments with longitudinal ultrasound waves.22-24 It has been found that relaxation mechanisms manifested in shear and bulk deformations are of a similar nature. In particular, polymeric solutions are characterized by close values of the temperature-shift factors and similar relaxation behavior of both shear and bulk viscosity. The data on the function, F3(λ), indicate that relaxation behavior of isotropic deformation at thermal expansion can be neglected for temperatures well above the glass-transition temperature.22 7.2.1.2 Dynamic interaction of bubbles with polymeric liquid Behavior of bubbles in liquid at varying external pressure or temperature is governed by cooperative action of a number of physical mechanisms, which are briefly discussed below. Sufficiently small bubbles execute radial motions (growth, collapse, pulsations) retaining their spherical shape and exchanging heat, mass and momentum with environment. Heat transfer between phases at free oscillations of gas bubbles is caused by gas heating during compression and its cooling when expanding. Due to the difference in thermal resistance of liquid and gas, the total heat flux from gas to liquid is positive for the oscillation period. This unbalanced heat exchange is the source of so-called heat dissipation. The magnitude of the latter depends on the relation between the natural time of the bubble (the Rayleigh time) t0=R0(ρf0/pf0)1/2, governed by the liquid inertia, and the time of temperature leveling in gas (characteristic time of heat transfer in a gas phase), tT = R02/ag, that is from the thermal Peklet number, PeT = tT/t0, where: R ag kg cgp 0 f g

radius of the bubble thermal diffusivity of gas, ag= kg/(pg0cgp) heat conductivity of gas specific heat capacity of gas at constant pressure index, referring to equilibrium state index, referring to liquid index, referring to gas

In the limiting cases PeT >> 1 and PeT > 1 and ω PeT t0 or PeDg > 1 where: tDg Dg PeDg

characteristic time of binary diffusion in vapor-gas phase, tDg= R02/Dg diffusion coefficient diffusion Peklet number for the vapor-gas phase, PeDg= tDg/t0

Fast motions of a bubble surface produce sound waves. Small (but non-zero) compressibility of the liquid is responsible for a finite velocity of acoustic signals propagation and leads to appearance of additional kind of the energy losses, called acoustic dissipation. When the bubble oscillates in a sound field, the acoustic losses entail an additional phase shift between the pressure in the incident wave and the interface motion. Since the bubbles are much more compressible than the surrounding liquid, the monopole sound scattering makes a major contribution to acoustic dissipation. The action of an incident wave on a bubble may be considered as spherically-symmetric for sound wavelengths in the liquid lf>>R0. When the spherical bubble with radius R0 is at rest in the liquid at ambient pressure, pf0, the internal pressure, pin, differs from pf0 by the value of capillary pressure, that is p in = p f 0 + 2σ / R0

[7.2.34]

where: σ

surface tension coefficient

If the system temperature is below the boiling point at the given pressure, pf0, the thermodynamic equilibrium of bubble in a liquid is possible only with a certain amount of inert gas inside the bubble. The pressure in vapor-gas mixture follows the Dalton law, that suggests that both the solvent vapor and the gas are perfect gases: p in = p g + pv = (ρ g B g + ρ v Bv )Tm = ρ m B mTm , ρ m = ρ g + ρ v

[7.2.35]

B m = (1 − k 0 )B g + k 0 Bv , B g,v = Rg / µ g,v where: k0 µ g,v v

equilibrium concentration of vapor inside the bubble molar masses of gas and solvent vapor index, referring to vapor

From [7.2.34], [7.2.35] follows the relation for k0:

[

{(1+ 2 σ) / p

k 0 = 1+ Bv Bg−1

v0

}]

−1

−1

,

pv 0 = pv 0 / p f 0 ,

σ = σ / ( pv 0R0 )

[7.2.36]

The equilibrium temperature enters equation [7.2.36] via the dependence of the saturated vapor pressure, pv0, from T0. Figure 7.2.1 illustrates the relation [7.2.36] for air-vapor bubbles in toluene.25 The curves 1- 3 correspond to temperatures T0 = 363, 378, 383.7K (the


365

latter value is equal to the saturation temperature Ts for toluene at pf0 = 105 Pa). It follows from the calculated data that the vapor content dependence on the bubble radius is manifested only for minor bubbles as a result of capillary forces. The effect vanish for R0> >10 mkm. For T01 mkm). Phase plot of this equation is presented in Figure 7.2.2. It is seen that for k = - 1 (collapsing cavity) z → z1 as t → ∞ if z0 > z2. The stationary point z = z2 is unstable. The rate of the cavity collapse z = z1 in the asymptotic regime satisfies inequality z p ≤ z1 ≤ 0, where zp = -Rep is equal to the collapse rate of the cavity in a pure viscous fluid with viscosity of polymeric solution η. It means that the cavity closure in viscoelastic solution of polymer at asymptotic stage is slower than in a viscous liquid with the same equilibrium viscosity. On the contrary, the expansion under the same conditions is faster: at k = 1 z p ≤ z1 ≤ z x , where zp = Rep and zs=Res =(1 - β)-1Rep is the asymptotic rate of the cavity expansion in a pure solvent with the viscosity (1- β)η. This result is explained by different behavior of the stress tensor component τrr, controlling the fluid rheology effect on the cav-


367

Figure 7.2.2. Phase plane for expanding and collapsing cavity in polymeric solution. [By permission of Nauka i Tekhnica Press, from the reference 25]

ity dynamics, in extensional and compressional flows, respectively.9 In the former case, the τrr value may be considerably greater than in the latter one. Heat transfer between phases is a strong dissipative factor that in principle can mask the rheological features in bubble dynamics. Nevertheless, even with account for heat dissipation the theoretical dependencies of R(t) are sensitive to rheological properties of solution. Typical results of air bubble dynamics simulations at a sudden pressure change in the solution with ηp >> ηs are presented on the Figure 7.2.3,

Figure 7.2.3. Heat transfer and rheodynamics at non-linear oscillations of a bubble in polymeric liquid. [By permission of Nauka i Tekhnica Press, from the reference 25]

where: R* τ τ*rr θ10 Q* kg ∆pf* h(t)

dimensionless radius of the bubble, R* = R/R0 dimensionless time, τ = t/t0 dimensionless radial component of the extra-stress tensor at the interface, τ*rr = τrr(R,t)/pf0 dimensionless temperature at the center of the bubble, θ10 = Tg(0,t)/T0 dimensionless heat, transferred to liquid from the gas phase in a time τ, Q* = Q/(R0T0kgt0) heat conductivity of gas dimensionless pressure change in the liquid at initial moment of time, pf*(∞) = 1 + ∆pf*h(t), ∆pf* = ∆pf/pf0 unit step function

Calculations have been done for the rheological model [7.2.25] with 20 relaxation elements in the spectrum, distributed according to the law [7.2.29] with z = 2. To illustrate the contribution of rheological non-linearity in equation [7.2.25] the numerical coefficient α (α=1 or 0) was introduced in the term with λk, containing material derivative. The value α=1 corresponds to non-linear model [7.2.25], while at α = 0 equation [7.2.25] is equivalent to the linear hereditary model [7.2.10] with a discrete spectrum. Other parameters of the system were chosen as follows: ηp = 2 Pas, ηs = 10-2 Pas, λ1 = 10-5 s, R0 = 50 mkm, ∆pf* = 10, pf0 = 105 Pa, ρf0 = 103 kg/m3 , T0 = 293K, σ = 0.05 N/m. Thermodynamic parameters of the

368


air were accepted according to the standard data,33 the heat transfer between phases was described within homobaric scheme34 (pressure in the bubble is a function of t only, that is uniform within the volume, while the density and temperature are changed with r and t according to the conservation laws). For the curve 1, ηp = ηs = 2 Pas (pure viscous liquid with Newtonian viscosity of the solution), for curve 2, α = 1, for curve 3, α = 0, that is, the latter two graphs correspond to viscoelastic solution with and without account for the rheological non-linearity, respectively. It follows from Figure 7.2.3 that relaxation properties of liquid are responsible for amplification of the bubble pulsations and, as a result, change the heat transfer between phases. Note that in examples, reproduced in Figure 7.2.3, the characteristic time of the pulsations’ damping is less than characteristic time of the temperature leveling in gas, tT (the time moment τ = 2.4, for instance, corresponds to t/tT ≈ 0.1). Therefore, after completion of oscillations the temperature in the center of a bubble is reasonably high, and the R* value exceeds the new isothermal equilibrium radius (R1* = 0.453). The manifestation of rheological non-linearity leads to a marked decrease in deviations of the bubble radius from the initial value at the corresponding instants of time. The explanation follows from stress dynamics analysis in liquid at the interface. At the initial stage, the relaxation of stresses in the liquid slows down the rise in the τ*rr value in comparison with similar Newtonian fluid. This leads to acceleration of the cavity compression. Since the stresses are small during this time interval, the rheological non-linearity has only a minor effect on the process. Further on, however, this effect becomes stronger which results in a considerable increase of normal stresses as compared with those predicted by the linear theory. It leads to deceleration of the cavity compression and, as a result, to decrease both in the maximum temperature of gas in a bubble and in the integral heat loss. More detailed information about rheological features in gas bubble dynamics in polymeric solutions can be received within linear approach to the same problem that is valid for small pressure variations in the liquid. The equation describing gas bubble dynamics in a liquid with rheological equation [7.2.10] follows from [7.2.41], [7.2.10] and has the form35 3γ  && 3 & 2  ρ f 0  RR + R  = p g 0 (R0 / R) − p f (∞) − 2σR −1 − 4ηs R / R& − 2   ∞ t  t − t ′  & ′ −1 ′ ′  −4∫ F1 (λ) ∫ exp −  R(t )R (t )dt  dλ λ   0 0 

[7.2.42]

Here is supposed that gas in the bubble follows polytropic process with exponent λ. This equation was solved in linear approximation35 by operational method with the aim to analyze small amplitude, natural oscillations of the constant mass bubble in relaxing liquid. It was taken R = R0 + ∆R, ∆R/R0 Ts(k0, pf0), assuming that both pressure and temperature in the vapor phase are constant pv = pf0, Tv = Ts(pf0, kR) = TfR, kR = k(R, t),

TfR = Tf(R, t)

[7.2.44]

Parameters k0, Tf0 characterize the state of solution far from the bubble (at r = ∞). Unlike a one-component liquid, the temperature TfR here is unknown. It is related to the surface concentration of solvent, kR, by the equation of phase equilibrium at the interface. Equations for heat transfer and diffusion in the solution have the form ∂T f ∂T f  ∂ R 2 ∂T f + v fR 2 = r −2 af r 2  ∂t ∂r ∂r  ∂r  r

[7.2.45]

∂k R 2 ∂k ∂  2 ∂k  + v fR 2 = r −2   Dr ∂t ∂r  ∂r  r ∂r

[7.2.46]


373

Since the thermal diffusivity of solution, af, is less affected20 by variations of temperature and concentration over the ranges Ts(kR) < Tf < Tf0 and kR < k < k0, respectively, than the binary diffusion coefficient, D, it is assumed henceforward that af = const. Furthermore, since the thermal boundary layer is much thicker than the diffusion layer, it is appropriate to assume that within the latter D = D(k,TfR). The boundary conditions for equations [7.2.45], [7.2.46] are as Tf = Tf0,

k = k0

at

r=∞

[7.2.47]

R& − v fR = ρ −f 1 j, R& = ρ v−1 j

[7.2.48]

(

)

j vfR kf ρf, ρv

phase transition rate per unit surface area of a bubble radial velocity of the liquid at the interface heat conductivity of liquid densities of solution and solvent vapor

∂k j = R& − v fR ρ f k R + ρ f D , ∂r

jl = k f

∂Tf ∂r

at r = R(t )

[7.2.49]

where:

Equations [7.2.48] and [7.2.49] yield j=

ρf ∂k |r = R D 1 − k R ∂r

[7.2.50]

If thermodynamic state of the system is far from the critical one, ε = ρ v / ρ f > 1, using the perturbation method:52

374

Semyon Levitsky, Zinoviy Shulman h = ( 6 / π )Ja 2 = ( 6π )Le −1Di 2 (1+ M1) , Di = ε−1Kα[1+ f (k R )], Kα = (k 0 − k R ) / (1− k R ) [7.2.54] 2

where: Ja ∆Tf Le Kα

Jacob number, Ja = cf∆Tf(εl)-1 superheat of the solution with respect to the interface, ∆Tf = Tf0 - TfR Lewis number, af /D0 mass fraction of the evaporated liquid46

Here M1 follows certain cumbersome equation,52 including f(k). The approximation Ja>>1 corresponds to the case of a thin thermal boundary layer around the growing bubble. Since, for polymeric solutions Le >> 1, the condition of small thickness of the diffusion boundary layer is satisfied in this situation as well. We start the analysis of the solution [7.2.54] from the approximation f = 0 that corresponds to D ≈ D0 = const. Then from [7.2.54] it follows: Kα =

( Le)c l f

−1

∆T

[7.2.55]

Because of the diffusion resistance, the solvent concentration at the interface is less then in the bulk, kR < k0. Writing the equation of phase equilibrium in linear approximation with respect to ∆k = k0 - k, from [7.2.55] one can receive49,53 −1

   ∂T  ∆T / ∆T * = 1− cf l −1(1− k R ) Le  s   ,  ∂k  k = k 0  

∆T * = Tf 0 − Ts (k 0 )

[7.2.56]

Here ∆T* represents the superheat of the solution at infinity. For solutions of polymers ∂Ts/∂k < 0 and, therefore, the actual superheat of the liquid ∆T < ∆T*. Additional simplification can be achieved if 1 - kR >> k0 - kR. It permits to assume in [7.2.56] kR ≈k0 and, hence, to find easily the vapor temperature. In the diffusion-equilibrium approximation (i.e. Le → 0) ∆T = ∆T*. When the diffusion resistance increases, the actual superheat ∆T lowers and, according to [7.2.56], at Le → ∞ ∆T → 0. However, in the latter case the assumptions made while deriving [7.2.56], are no longer valid. Indeed, the Ja number, connected with the superheat of the solution with respect to the interface, is related to the Ja0 value, corresponding to the bulk superheat, by Ja=Ja0(∆T/∆T*). Since the ratio ∆T/∆T* varies in the range (0, 1), then, at small diffusion coefficients, it may be that Ja > 1. In this case, the asymptotic solution of the problem takes the form46 h = Ja, and, for thin diffusion boundary layer, it can be received instead of [7.2.54]: h = Ja = (6 / π)Le −1Di 2 (1 + M1 )

[7.2.57]

Finally, at Di > 1 by increasing the superheat because of low values of the corresponding Sn numbers. 7.2.3 BOILING OF MACROMOLECULAR LIQUIDS Experimental investigations of heat transfer at boiling of polymeric liquids cover highly diluted (c = 15 to 500 ppm), low-concentrated (c ~ 1%), and concentrated solutions (c>10%). The data represent diversity of physical mechanisms that reveal themselves in boiling processes. The relative contribution of different physical factors can vary significantly with changes in concentration, temperature, external conditions, etc., even for polymers of the same type and approximately equal molecular mass. For dilute solutions this is clearly demonstrated by the experimentally detected both intensification of heat transfer at nucleate boiling and the opposite effect, viz. a decrease in the heat removal rate in comparison with a pure solvent. Macroscopic effects at boiling are associated with changes in the intrinsic characteristics of the process (e.g., bubble shape and sizes, nucleation frequency, etc.). Let’s discuss the existing experimental data in more detail. One of the first studies on the effect of water-soluble polymeric additives on boiling was reported elsewhere.56 For a plane heating element a significant increase in heat flux at fixed superheat, ∆T = 10-35K, was found in aqueous solutions of PAA Separan NP10 (M = 106), NP20 (M = 2×106), and HEC (M ~ 7×104 to about 105) at concentrations of 65 to 500 ppm (Figure 7.2.13). The experiments were performed at atmospheric pressure; the viscosity of the solutions did not exceed 3.57×10-3 Pas. The following specific features of boiling of polymer solution were revealed by visual observations: (i) reduction in the departure diameter of bubbles, (ii) more uniform bubFigure 7.2.13. Effect of the HEC additives on the boil- ble-size distribution, (iii) decrease in the ing curve. 1 - pure water; 2, 3 and 4 - HEC solution with tendency to coalescence between bubbles. c = 62.5, 125 and 250 ppm, correspondingly. [Reprinted The addition of HEC led to faster covering from P. Kotchaphakdee, and M.C. Williams, Int. J. Heat Mass Transfer, 13, 835, Copyright 1970, the reference of the heating surface by bubbles during the 52, with permission from Elsevier Science] initial period of boiling and bubbles were

378

Figure 7.2.14. The relation between the relative heat transfer coefficient for boiling PIB solutions in cyclohexane and the Newtonian viscosity of the solutions measured at T=298 K. ∆T = 16.67 K; l - PIB Vistanex L-100 in cyclohexane, o - PIB Vistanex L-80 in cyclohexane, x - pure cyclohexane. [Reprinted from H.J. Gannett, and M.C. Williams, Int. J. Heat Mass Transfer, 14, 1001, Copyright 1971, the reference 57, with permission from Elsevier Science]


Figure 7.2.15. The average bubble detachment diameter in boiling dilute aqueous solutions of PEO.59 ∆T = 15K. For curves 1-3 the flow velocity v = 0, 5×10-2, and 10-1 m/s, respectively. [Adapted, from S.P. Levitsky, and Z.P. Shulman, Bubbles in polymeric liquids, Technomic Publish. Co., Lancaster, 1995, with permission from Technomic Publishing Co., Inc., copyright 1995]

smaller in size than in water and aqueous solutions of PAA. Non-monotonous change in the heat transfer coefficient, α, with increasing the concentration of PIB Vistanex L80 (M = 7.2×105) or L100 (M = 1.4×106) in boiling cyclohexane has been reported.57 The results were received in a setup similar to that described earlier.56 It was found that the value of α increases with c in the range 22 ppm < c < 300 ppm and decreases in the range 300 ppm < c < 5150 ppm. Viscosity of the solution, corresponding to αmax value, according to the data57 only slightly exceeds that of the solvent (Figure 7.2.14). Within the entire range of concentrations at supercritical (with respect to pure solvent) superheats, the film boiling regime did not appear up to the maximum attainable value ∆T ~ 60K. The growth of polymer concentration in the region of “delayed ” nucleate boiling led to a considerable decrease in heat transfer. These findings56,57 were confirmed58 in a study of the nucleate boiling of aqueous solutions of HEC Natrosol 250HR (M = 2×105), 250GR (M = 7×104), and PEO (M ~ (2-4)×106) at forced convection of the liquid in a tube. A decrease in the size of bubbles in the solution and reduction of coalescence intensity were recognized. Similar results were presented also in study,59 where the increase in heat transfer at boiling of aqueous solutions of PEO WSR-301 (M=2×106) and PAA Separan AP-30 (15 ppm < c < 150 ppm) on the surface of a conical heater was observed. In aqueous solutions of PAA with c > 60 ppm the α value began to decrease. With an increase in c the detachment diameter of bubbles decreased (Figure 7.2.15), the nucleation frequency increased, and the tendency to coalescence was suppressed. Boiling of PEO solutions with c = 0.002 to 1.28% at atmospheric and sub-atmospheric pressures was examined60 for subcoolings in the range 0 to 80K. It was demonstrated that at saturated boiling the dependence of the heat transfer coefficient α on the polymer concentration is non-monotonous: as c grows, α first increases, attaining the maximal value at c≈0.04% , whereas at c = 1.28% the value of α is smaller than in water (α < αs) (Figure 7.2.10). With a decrease in pressure the effect of polymeric additives weakens and for solution with greatest PEO concentration (in the investigated range) the α value increases, approaching αs from below. The critical heat flux densities in PEO solutions are smaller than those for water.


379

In view of the discussed results, the work61 attracts special attention since it contains data on boiling of dilute solutions, opposite to those reported earlier.56-60 The addition of PAA, PEO and HEC to water in concentrations, corresponding to the viscosity increase up to ηp = 1.32×10-3 Pas, has brought about reduction in heat transfer. The boiling curve in coordinates q (heat flux) vs. ∆T displaced almost congruently to the region of larger ∆T values with c (Figure 7.2.17a). It was demonstrated61 that the observed decrease in α with addition of polymer to water can be both qualitatively and quantitatively (with the Rohsenow pool boiling correlation for the heat transfer coefficient62) associated with the increase in the solution viscosity (Figure 7.2.17, b). The experiments61 were performed using a thin platinum wire with diameter 0.3 mm. Explanation of experimental data needs more detailed discussion of physical factors that can reveal themselves in boiling of polymeric solutions. They include possible changes in capillary forces on interfaces in the presence of polymeric additives; absorption of macromolecules on the heating surface; increase in the number of weak points in the solution, which facilitates increase in the number of nuclei; thermodynamic peculiarities of the polymer-solvent system; the effect of macromolecules on the diffusion mass transfer in evaporation of solvent; hydrodynamics of convective flows in a boiling layer and the motion of bubbles; manifestation of rheological properties of solution. The capillary effects were indicated as one of the reasons for the intensification of heat transfer, since many polymers (in particular, HEC, PEO, etc.),63 similar to low-molecular surfactants,64 are capable of decreasing the surface tension. As a result, Figure 7.2.16. Heat transfer coefficient for nucleate pool they decrease both the work of the nucleus boiling of PEO aqueous solutions. (pf0 = 9.8×103 Pa). formation, Wcr, and the critical size of bubCurve 1 corresponds to pure water, for curves 2 - 7, c = ble, Rcr: 0.01, 0.02, 0.04, 0.08, 0.16 and 1.28%, respectively. [Adapted, from S.P. Levitsky, and Z.P. Shulman, Bubbles in polymeric liquids, Technomic Publish. Co., Lancaster, 1995, with permission from Technomic Publishing Co., Inc., copyright 1995]

Figure 7.2.17. Boiling curves for aqueous solutions of PAA Separan AP-30. (a) experimental data; for curves 1 - 6 ηr = 1.00, 1.01, 1.04, 1.08, 1.16 and 1.32, correspondingly; (b) calculations made with the use of the Rohsenow pool boiling correlation; for curves 1 - 5, ηr = 1.00, 1.01, 1.04, 1.16 and 1.32, respectively (ηr = ηp/ηs). [By permission from D.D. Paul, and S.I. Abdel-Khalik, J. Rheol., 27, 59, 1983, reference 61]

380

Semyon Levitsky, Zinoviy Shulman W cr = 16 / 3πσ 3 Φ( θ)[(dp / dT )∆T (1− ρ v / ρ f )] , 2

Rcr = 2σ[(dp / dT )∆T (1 − ρ v / ρ f )]

Φ( θ) = 1/ 4( 2 + 3 cos θ − cos 3 θ) [7.2.59]

−1

where: θ σ

wetting angle surface tension coefficient

However, it should be noted that the integral effect of the heat transfer enhancement, observed in highly diluted solutions, can not be attributed to the capillary phenomena alone, since the main change in σ occurs in the range of low polymer concentrations59 (c < 50 ppm) and further increase in c does not affect the value of σ, whereas the α value continues to grow. PAA, for example, does not behave like surfactants at all. It should be noted also that in the presence of polymer not only the value of σ changes, but also the wetting angle, θ, in the formula [7.2.59]. The latter may lead to manifestation of different behavior. Absorption of macromolecules onto a heating surface favors the formation of new centers of nucleation. Together with an increase in nucleation sites in the boundary layer of a boiling liquid it explains the general growth in the number of bubbles. Both this factor and reduction in the σ value for solutions of polymers that possess surface activity, are responsible for a certain decrease in superheat needed for the onset of boiling of dilute solutions.57,60 The decrease in the water vapor pressure due to presence of polymer in solution at c~1% can be neglected. However, if the solution has the LCST, located below the heating wall temperature, the separation into rich-in-polymer and poor-in-polymer phases occurs in the wall boundary layer. At low concentration of macromolecules the first of these exists in a fine-dispersed state that was observed, for example, for PEO solutions.60 The rich-in-polymer phase manifests itself in a local buildup of the saturation temperature, which can be significant at high polymer content after separation; in decrease of intensity of both convective heat transfer and motion of bubbles because of the increase in viscosity; and reduction of the bubble growth rate. The so-called “slow” crisis, observed in PEO solutions60 is explained by integral action of these reasons. Similar phenomenon, but less pronounced, was observed also at high enough polymer concentrations.58 It is characterized by plateau on the boiling curves for solutions of PIB in cyclohexane, extending into the range of high superheats. The main reason for the decrease in heat transfer coefficient at nucleate boiling of polymeric solutions with c ~ 1% is the increase in liquid viscosity, leading to suppression of microconvection and increasing the resistance to the bubbles’ rising. In the presence of LCST, located below the boiling temperature, the role of this factor increases because appearance in the boiling layer of the rich-in-polymer phase in fine-dispersed state. Another reason for the decrease of α in the discussed concentration range ∧is the decrease in the bubble growth rate at the thermal stage, when the superheat ∆T > ∆ T* (Section 7.2.2). In highly diluted solutions the change in Newtonian viscosity due to polymer is insignificant, and though the correlation between heat transfer enhancement and increase in viscosity has been noticed, it cannot be the reason for observed changes of α. In hydrodynamics, the effect of turbulence suppression by small polymeric additives is known, but it also cannot be considered for such a reason because laminarization of the boundary layer leads to reduction of the intensity of convective heat transfer.65 Nevertheless, the phenomenon of the decrease of hydrodynamic resistance and enhancement of heat transfer in boiling dilute solutions have a common nature. The latter effect was connected


381

Figure 7.2.18. Growth of vapor bubbles on the heating surface at high (a) and low (b) pressures. [Reprinted from S.P. Levitsky, B.M. Khusid and Z.P. Shulman, Int. J. Heat Mass Transfer, 39, 639, Copyright 1996, the reference 66, with permission from Elsevier Science]

with manifestation of elastic properties of the solution at vapor bubble growth on the heating surface.66 The general character of the bubbles evolution at boiling under atmospheric and subatmospheric pressures, respectively, is clarified schematically in Figure 7.2.18. In the first case (at high pressures) the base of a bubble does not “spread”67 but stays at the place of its nucleation. Under such conditions the decrease in the curvature of the bubble surface with time, resulting from the increase in bubble radius, R, leads to liquid displacement from the zone between the lower part of the microbubble and the heating surface. This gives rise to the local shear in a thin layer of a polymer solution. A similar shear flow is developed also in the second case (at low pressures), when a microlayer of liquid is formed under a semi-spherical bubble. As known, at shear of a viscoelastic fluid appear not only tangential but also normal stresses, reflecting accumulation of elastic energy in the strained layer (the Weissenberg effect3). The appearance of these stresses and elastic return of the liquid to the bubble nucleation center is the reason for more early detachment of the bubble from the heating surface, reduction in its size and growth in the nucleation frequency. All this ultimately leads to enhancement of the heat transfer. The above discussion permits to explain the experimental results.61 Their reasons are associated with substantial differences in the conditions of boiling on a thin wire and a plate or a tube. Steam bubbles growing on a wire have a size commensurable with the wire diameter (the growing bubble enveloped the wire61). This results in sharp reduction of the boundary layer role, the same as the role of the normal stresses. Besides, the bubble growth rate on a wire is smaller than on a plane (for a wire R ~ tn where n < 1/4).67 The elastic properties of the solution are responsible also for stabilization of the spherical shape of bubbles observed in experiments on boiling and cavitation. Finally, the observed reduction in a coalescence tendency and an increase in the bubble sizes uniformity can also be attributed to the effects of normal stresses and longitudinal viscosity in thin films separating the drawing together bubbles. The linkage between the enhancement of heat transfer at boiling of dilute polymer solutions and the elastic properties of the system is confirmed by the existence of the optimal concentration corresponding to αmax (Figure 7.2.14). Similar optimal concentration was established in addition of polymers to water to suppress turbulence - the phenomenon that also owes its origin to elasticity of macromolecules.1,3,9 Therefore, it is possible to expect that the factors favoring the chain flexibility and increase in the molecular mass, should lead to strengthening of the effect. The data on boiling of concentrated polymeric solutions20 demonstrate that in such systems thermodynamic, diffusional, and rheological factors are of primary importance.

382


The diagram of the liquid-vapor phase equilibrium is characterized by a decrease in the derivative dp/dT with the polymer concentration (dp/dT → 0 at k → 0). This leads to increase in both the nucleation energy and the detachment size of a bubble (equation [7.2.59]) and, consequently, to reduction of the bubbles generation frequency. Note that in reality the critical work, Wcr, for a polymeric liquid may exceed the value predicted by the formula [7.2.59] because of manifestation of the elasticity of macromolecules. As known,62 the heat transfer coefficient in the case of developed nucleate boiling of low-molecular liquids is related to the heat flux, q, by the expression α = Aqn where n≈0.6-0.7. For concentrated polymeric solutions the exponent n is close to zero. The decrease in heat transfer is explained by the increase in the viscosity of the solution near the heating surface, resulting from the evacuation of the solvent with vapor. Another reason for the decrease of α in such systems is the reduction of the bubble growth rate with lowering k0 and the impossibility to achieve large Ja numbers by rising the solution superheat. Since in boiling of concentrated polymer solutions the α value is small, the superheat of the wall at a fixed q increases. This can give rise to undesirable phenomena such as burning fast to the heating surface, structure formation, and thermal decomposition. Usually, in this case the heat transfer is intensified by mechanical agitation. Note that one of the promising trends in this field may become the use of ultrasound, the efficiency of which should be evaluated with account for considerable reduction in real losses at acoustically induced flows and pulsations of bubbles in viscoelastic media.68,69 Specific features of boiling of high-molecular solutions are important for a number of applications. One of examples is the heat treatment of metals, where polymeric liquids find expanding employment. The shortcomings of traditionally used quenching liquids, such as water and oil, are well known.70 Quenching in oil, due to its large viscosity and high boiling temperature, does not permit to suppress the perlite transformation in steels. From the other hand, water as a quenching medium is characterized by high cooling rate over the temperature ranges of both perlite and beinite transformations. However, its maximum quenching ability lies in the temperature range of martensite formation that can lead to cracking and shape distortion of a steel article. Besides, the quenching oils, ensuring the so-called “soft” quenching, are fire-hazardous and have ecological limitations. The polymeric solutions in a certain range of their physical properties combine good points of both oil and water as quenching liquids and permit to control the cooling process over wide ranges of the process parameters. For the aims of heat treatment a number of water-soluble polymers are used, e.g. PVA, PEO, PAA, polymethacrylic acids (PMAA, PAA) and their salts, cellulose compounds, etc.71-73 The optimal concentration range is 1 to 40% depending on molecular mass, chemical composition, etc. Typical 73 Figure 7.2.19. Cooling curves for a silver specimen data are presented in Figure 7.2.19, where o quenched in a polymer aqueous solution at 20 C. curves 1-5 correspond to the solution viscos[Adapted, from S.P. Levitsky, and Z.P. Shulman, Bub- ity η = (1.25, 2.25, 3.25, 5, 11)×10-3 Pas, p bles in polymeric liquids, Technomic Publish. Co., o Lancaster, 1995, with permission from Technomic measured at 40 C. The quenchants based on water-soluble polymers sustain high cooling Publishing Co., Inc., copyright 1995]


383

Figure 7.2.20. Stability diagrams for film boiling.73 Quenching in water (a) and in aqueous polymer solution with ηs = 3.27×10-3 Pas at 40oC; (b): 1, 2 - stable and unstable film boiling, 3 - nucleate boiling, 4 - convection. Tms, Tm1 and Tm2 are the surface temperatures of the specimen, corresponding to the destabilization of the regimes 1 - 3, respectively. [Adapted, from S.P. Levitsky, and Z.P. Shulman, Bubbles in polymeric liquids, Technomic Publish. Co., Lancaster, 1995, with permission from Technomic Publishing Co., Inc., copyright 1995]

rates during the initial stage of quenching that permits to obtain fine-grained supercooled austenite, and relatively low intensity of heat removal at moderate temperatures, when martensite transformations take place. The main feature of quenching in polymeric solutions is the prolongation of the cooling period as a whole in comparison with water that is explained by extended range of a stable film boiling (Figure 7.2.20). The increase in polymer concentration leads to reduction of α on the stage of nucleate boiling and growth of the temperature, corresponding to the onset of free convection regime. The effect of polymeric additives on the initial stage of the process was the subject of a special investigation.74 Experiments were performed with aqueous polymer solutions of Breox, PEO and some other polymers with M = 6×103 to 6×105 at pressure of 0.1 MPa. The platinum heater with short time lag, submerged in solution, was heated in a pulsed regime & ~ 105 to 106 K/s. The experimental results revealed the existence of a period with enwith T hanced heat transfer (as compared to water) in solutions with c ~ 1%, which lasted for 10 to 100 µs after the onset of ebullition. The sensitivity of heat transfer to the polymer concentration was sufficiently high. After formation of the vapor film the secondary ebullition was observed, which resulted from superheating of the liquid outside the region of concentration gradients near the interface. The mechanism of this phenomena was described.75 It is associated with the fact that heat transfer has a shorter time lag than mass transfer, and thus the thermal boundary layer in a liquid grows faster than the diffusion one. The experimental data and theoretical results on the growth of vapor bubbles and films in polymeric solutions explain the efficiency of quenchants, based on water-soluble polymers. The main reason is stabilization of the film-boiling regime at initial stage of quenching. Such stabilization is connected with elastic properties of the liquid skin layer, adjacent to the interface that is enriched by polymer due to solvent evaporation. Appearance of this layer leads to fast growth of longitudinal viscosity and normal stresses, when perturbations of the interface arise, thus increasing the vapor film stability. A similar mechanism is responsible for stabilization of jets of polymeric solutions9 as well as for retardation of bubble

384


collapse in a viscoelastic liquid (Section 7.2.1). In systems with LCST this effect can be still enhanced due to phase separation, induced by interaction of the liquid with high-temperature body. Additional reason that governs the decrease in heat removal rate at quenching is connected with reduction of heat conductivity in aqueous solutions of polymers with growth of concentration of the high-molecular additive. After the body temperature is lowered sufficiently, the film boiling gave way to the nucleate one. According to data, presented in the previous section, over the polymer concentration ranges, typical for high-molecular quenchants, the α value must decrease in comparison with water. In fact, this is normally observed in experiments. Rise of the temperature, characterizing transition from nucleate boiling to convective heat transfer, is associated with the increase in liquid viscosity. Abbreviations HEC LCST PS PIB PAA PEO POE PVA TTS

hydroxyethylcellulose lower critical solution temperature polystyrene polyisobutylene polyacrylamide polymethyleneoxide polyoxyethylene polyvinyl alcohol time-temperature superposition

REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

A.B.Bird, R.C.Armstrong, and O.Hassager, Dynamics of polymeric liquids, J.Wiley & Sons, New York, 1977. R. M.Christensen, Theory of viscoelasticity. An introduction, Academic Press, New York, 1982. G. Astarita, and G. Marucci, Principles of non-Newtonian fluid mechanics, McGraw-Hill Book Co., London, 1974. L. D. Landau, and E. M. Lifshits, Statistical physics, Pergamon Press, Oxford, 1980. C. Truesdell, A first course in rational continuum mechanics, Johns Hopkins University, Baltimore, 1972. K. Walters, Rheometry, Chapman and Hall, London, 1975. C.J.S. Petrie, Elongational flows, Pitman, London, 1979. P. J. Carreau, and D. De Kee, Can. J. Chem. Eng., 57, 3 (1979). Z.P. Shulman, and B. M. Khusid, Non-stationary convective transfer processes in hereditary media, Nauka i Technika, Minsk, 1983. K.M. Baid, and A. B. Metzner, Trans. Soc. Rheol., 21, 237 (1977). R.Y. Ting, J. Appl. Polym. Sci., 20, 1231 (1976). J.M. Dealy, Polym. Eng. Sci., 11, 433 (1971). M.M. Denn, and G. Marrucci, AIChE J., 17, 101 (1971). J.D. Ferry, Viscoelastic properties of polymers, J.Wiley & Sons, New York, 1980. M. Doi, and S. F. Edwards, The theory of polymer dynamics, Clarendon Press, Oxford, 1986. P.-G. de Gennes, Scaling concepts in polymer physics, Cornell University Press, Ithaca and London, 1979. T. W. Spriggs, Chem. Eng. Sci., 20, 931, (1965). G. V. Vinogradov, and A. Ya. Malkin, Rheology of polymers, MIR, Moscow, 1980. Ya.I. Frenkel, Kinetic theory of liquids, Dover Publications, New York, 1955. V.P. Budtov, and V.V. Konsetov, Heat and mass transfer in polymerization processes, Khimiya, Leningrad, 1983. D. W. Van Krevelen, Properties of polymers: correlations with chemical structure, Elsevier Publishing Co., Amsterdam, 1972. R. S. Marvin, and J. E. McKinney in Physical Acoustics: Principles and Methods, vol.II, p.B, W. P. Mason, Ed., Academic Press, New York, 1965, pp. 193-265. B. Froelichb, C. Noelb, and L. Monnerie, Polymer, 20, 529 (1979). D. Pugh, and D.A. Jones, Polymer, 19, 1008 (1978).


25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75

385

S.P. Levitsky, and Z.P. Shulman, Dynamics and heat and mass transfer of bubbles in polymeric liquids, Nauka i Tekhnika, Minsk, 1990. W.J. Yang, and H.C. Yeh, Phys. Fluids., 8, 758 (1965). H.S. Fogler, and J.D. Goddard, Phys. Fluids., 13, 1135 (1970). J.R. Street, A.L. Fricke, and L.P. Reiss, Ind. Eng. Chem. Fundam., 10, 54 (1971). S.P. Levitsky, and A.T. Listrov, J. Appl. Mech. Techn. Phys., 15, 111 (1974). G. Pearson, and S. Middleman, AIChE J., 23, 714 (1977). S.P. Levitsky, J. Appl. Mech. Techn. Phys., 20, 74 (1979). G. Ryskin, J. Fluid Mech., 218, 239 (1990). R.C. Reid, J.M.Prausnitz, and T.K. Sherwood, The properties of gases and liquids, McGraw-Hill, New York, 1977. R.I. Nigmatulin, and N.S. Khabeev, Fluid Dynamics, 9, 759 (1974). S.P. Levitsky, and Z.P. Shulman, Bubbles in polymeric liquids, Technomic Publish. Co., Lancaster, 1995. R.Y. Ting, and A.T. Ellis, Phys. Fluids, 17, 1461 (1974). G.L. Chahine, and D.H. Fruman, Phys. Fluids, 22, 1406 (1979). A. Shima, Y. Tomito, and T. Ohno, Phys. Fluids, 27, 539 (1984). W.D. McComb, and S. Ayyash, J. Phys. D: Appl. Phys., 13, 773 (1980). S.L. Peshkovsky, M.L. Fridman, V.I. Brizitsky et al., Doklady Akad. Nauk SSSR, 258, 706 (1981). R.T. Knapp, J.W. Daily, and F.G. Hammit, Cavitation, McGraw-Hill, New York, 1970. P.R. Williams, P.M. Williams, and S.W. Brown, J. Non-Newtonian Fluid Mech., 76, 307 (1998). J.W. Hoyt, Trans. ASME. J. Fluids Eng., 98, 106 (1976). R.Y. Ting, AIChE J., 20, 827 (1974). R.Y. Ting, Phys. Fluids, 21, 898 (1978). L.E. Scriven, Chem. Eng. Sci., 10, 1 (1959). P.J. Bruijn, Physica, 26, 326 (1960). R.A. Shock, in Multiphase science and technology, Hemisphere Publishing Corporation, New York, 1981, pp. 281-386. J.R. Thome, and R. A. W. Shock, Adv. Heat Transfer, 16, 60 (1984). S.G. Kandlikar, Trans. ASME., J. of Heat Transfer, 120, 380 (1998). J. Crank, The mathematics of diffusion, Clarendon Press, Oxford, 1975. Z.P. Shulman, and S.P. Levitsky, Int. J. Heat Mass Transfer, 39, 631 (1996). L.W. Florshuets, and A. R. Khan, Heat Transfer-70, v.6, p. B7, 31970, Paris, 1970, pp.1-11. R.I. Nigmatulin, Dynamics of multiphase flow, Hemisphere Publishing Corporation, New York, 1990. J.L. Duda, J.S. Vrentas, S.T. Ju, and H.T. Liu, AIChE J., 28, 279 (1982). P. Kotchaphakdee, and M.C. Williams, Int. J. Heat Mass Transfer, 13, 835 (1970). H.J. Gannett, and M.C. Williams, Int. J. Heat Mass Transfer, 14, 1001 (1971). H. Wei, and J.R. Maa, Int. J. Heat Mass Transfer, 25, 431 (1982). A.T. Papaioannou, and N.G. Koumoutsos, 7th Int. Heat Transfer Conf. Proc., 1982, Munchen, 1982, v.4, pp. 67-72. B.P. Avksent’yuk, and Z.S. Mesarkishvili, Boiling of aqueous PEO solutions at reduced pressures under the conditions of natural convection, Institute of Thermophysics, RAS, Novosibirsk, 1983, preprint No. 108. D.D. Paul, and S.I. Abdel-Khalik, J. Rheol., 27, 59 (1983). Heat exchanger design handbook, v.1, contributors D.B. Spalding and J. Taborek, Hemisphere Publishing Corporation, New York, 1987. R.Y.Z. Hu, A.T.A. Wang, and J.P. Hartnett, Exp. Therm. Fluid Sci., 4, 723 (1991). Yan Yu Min, Int. Commun. Heat and Mass Transfer, 17, 711 (1990). Z.P. Shulman, B.M. Khusid, and S.P. Levitsky, Heat Transfer Research, 25, 872 (1996). S.P. Levitsky, B.M. Khusid and Z.P. Shulman, Int. J. Heat Mass Transfer, 39, 639 (1996). M.G. Cooper, and A.J.P. Lloyd, Int. J. Heat & Mass Transfer, 12, 895 (1969). Z.P. Shulman, and S.P. Levitsky, Int. J. Heat & Mass Transfer, 35, 1077 (1992). S.P. Levitsky, and Z.P. Shulman, Soviet Phys. Acoust., 31, 208 (1985). N.R. Suttie, Heat Treat. Metals, 6, 19 (1979). K.Y. Mason, and T. Griffin, Heat Treat. Metals, 9, 77 (1982). F. Moreaux, and G. Beck, Heat Transfer, 4, 2067 (1986). F. Morou, and P. Arshambol, Prom. Teplotekhn., 11, 48 (1989). G.B.Okonishnikov, N.V. Novikov, P.A. Pavlov, and S.L. Tsukrov, Heat and Mass Transfer - 92, v.6, Heat and Mass Transfer Institute, Minsk, 1992, pp. 3-6. S.P. Levitsky, and Z.P.Shulman, Thermophys. of High Temperatures, 33, 616 (1995).

386

Seung Su Kim, Jae Chun Hyun

7.3 DRYING OF COATED FILM Seung Su Kim SKC Co., Ltd., Chon-an City, Korea

Jae Chun Hyun Department of Chemical Engineering, Korea University, Seoul, Korea

7.3.1 INTRODUCTION Thin film coating and drying technology are the key technologies for manufacturing diverse kinds of functional films, such as photographic films, adhesives, image media, magnetic media and recently lithium battery coating. Coating applied to a substrate as a liquid need some degree of solidification in order to be final products. The degree of solidification can be low in the case of pressure-sensitive adhesives (PSA) and it ranges to high in the case of dense metal-oxides.48 The final structure and properties of coating are greatly influenced by the drying conditions.21 Poorly chosen operating conditions of drying cause unwanted internal gradients,6,7 phase separations,29,35,36 colloidal transformations that lead to the wrong microstructure,43,51,56 inappropriate non-uniformities, and stress-related defects.12-15,21,22 Here we address a subject of solvent removal, or drying, which is a part of solidification processes. Typically a formulation of coating solution is consisted of pigments, binders (polymer resins) and solvents. The solvents are used to solubilize the coating formulation and to give the coating solution(or dispersion) the rheology necessary for the application. The coating solution is deposited onto a substrate or web at the coating station and is dried by passing through a series of separate ovens (zones). A substrate can be an impermeable material such as plastic film and permeable in the case of paper coating. The dryer is consisted of ovens

Figure 7.3.1. An example of industrial coating and drying apparatus. A coated liquid is deposited onto a substrate which is unwound from a supply roll at the coating station and passes through the three separate ovens(dryer) where the temperature and velocity of air is controlled independently. Finally dried coated substrate is taken up by a take-up roll.

7.3 Drying of coated film

387

(zones) in which the temperature and velocity of air are controlled independently. The air impinges the coated and back side surface of the substrate through the nozzles and sweeps away the solvent vapor from the coated surface. In a case of single sided impingement dryer, the air impinges only on the coated surface. The coated film must be dry before the rewind station. The residence time of coated substrate in the dryer is as short as several seconds in a case of high speed magnetic coating processes and it ranges to as long as several minutes in a case of lithium battery coating processes.28 Finally the dried coated substrate is taken up by a take-up roll (Figure 7.3.1). The elemental process of drying is depicted in the Figure 7.3.2. Solvent is evaporated from the exposed surface of a liquid coating into the adjacent air. Diffusion of evaporating solvent into stagnant air is a comparatively slow process. Commonly the rate of diffusion of solvent vapor is greatly enhanced by the forced convective sweeping of the exposed surface. The rate of solvent evaporation per unit area is a product of the two factors : 1) the difference in partial pressure of the solvent at the surface of coating and in the bulk of nearby gas and 2) the mass transfer coefficient, which represents the combined action of convection and diffusion.16 The energy, which is needed to supply Figure 7.3.2. Elemental process of drying and typical parameters of drying. [After references 28,30]. the latent heat of solvent evaporation, is delivered mainly by blowing hot air onto the coating surface. This convective heat transfer not only can deliver the needed energy to the coating, but also can enhance the transport of solvent vapor away from the surface of the coating. Commonly the conductive and radiative heating are accompanied with convective heating if they are necessary. As long as the temperature and the concentration of solvent at the exposed surface of coating is constant, so does the evaporation rate of solvent. This is true during the initial stages of drying when the exposed surface of coating is fully wetted with the solvent. This period is commonly called as constant drying rate period (CDRP).10 In CDRP, all the heat, which is supplied to the coating, is used to supply latent heat of vaporization. Thus the temperature of coating surface is nearly constant. In a case of aqueous coating, the temperature of coating surface is equivalent to the wet bulb temperature of a given air humidity.9,10 Therefore in CDRP the external mass transfer resistance to drying limits the rate of drying. As the solvents depart the coating, the rate controlling step for the drying steadily shifts from the external mass transfer to the mass transfer within the coating. The solvents within the coating can move to the exposed surface by diffusion along with diffusion-induced convection1,8,17 and pressure gradient-driven flow in a porous coating.43,44 In polymeric solutions, the diffusion coefficient of solvent dramatically drops as solvent concentration falls.17 Thus the concentration of solvent at the exposed coating surface falls with drying proceeded, and so does the vapor pressure of solvent there. Therefore the drying

388


rate steadily falls. This period is called as falling drying rate period (FDRP). Generally the diffusion coefficient of solvent in a polymeric solutions is a strong function of solvent concentration, temperature and molecular size.53-55,60 The binary diffusion coefficient can be estimated by the free volume theory of Vrentas and Duda17 and can be measured by the NMR. However, there are many difficulties to measure or estimate the ternary diffusion coefficient. The evaporated solvents are swept away by the hot air and are exhausted to the outside of the dryer. The vapor concentration of solvent of a zone (oven) must be lower than the lower explosive level (LEL) of solvents to meet the safety. Thus enough fresh air should be supplied to each zone to meet this LEL safety.10,28 However, too much supply of fresh air brings about too much energy consumption to heat the fresh air, and it also increases the VOC containing waste gas which should be treated by VOC emission control unit. The LEL of a zone can be controlled within appropriate LEL ranges by measuring and predicting of drying rate.28 We are going to deal with some important features of drying a liquid coated film in the subsequent sections. First, we will introduce theory for the drying and method of modeling of the drying process. A drying process of liquid coated film can be simulated by setting up heat and mass transfer equations8,9,40,59 and vapor liquid equilibrium equations19,39 with appropriate heat and mass transfer coefficient.37,45 Secondly, we will try to give examples of drying rate measurements in laboratory experiments and in a pilot or production scale dryer. Traditionally, the rate of drying was obtained by measuring weight loss using a balance under the controlled experimental conditions of air velocity and temperature.5,40,51 However, it is difficult to obtain such data in the high airflow rate experiments, for the balance is disturbed by the high airflow motion. Therefore the alternative way of measuring was proposed in which hydrocarbon analyzers57 or FT-IR47,50 was used instead of balance. Moreover, the weight changes of each solvent in a multi-solvent coating system could be found by applying rapid scanning FT-IR method.50 However, rare data has been given about measuring the actual drying rate and solvent concentration profile along the dryer while operating.28 The drying rate of coated film in an industrial dryer is measured by analyzing dryer exit gas, and we will illustrate how the solvent concentration profiles of coating at the exit of each zone are estimated from the experimentally measured drying rate data.28 In third, we will try to illustrate the dependence of dried coating structure on the drying path by means of the basic phenomena, phase equilibrium.29,35,38,41,42,56,58 We will also briefly introduce the mechanisms behind the formation of microstructure of various kind of functional films.44,51 Finally, we will categorize the drying related defects according to the origin of defects, and we will show some examples of defects and the cause and curing of them.8,15,21,25 7.3.2 THEORY FOR THE DRYING 7.3.2.1 Simultaneous heat and mass transfer Because evaporation is an endothermic process, heat must be delivered to the system, either through convection, conduction, radiation, or a combination of these methods. The solvents are evaporated from the coating surface and at the same time the latent heat of solvent cool


389

down the coating surface. Thus the heat and mass are transferred simultaneously through the surface of coating. Assuming that the heat is supplied only by convection of hot air and the substrate is impermeable. Further if we neglect the internal resistance of solvent transport to the coating surface, the heat and mass balance consist a lumped parameter system. A schematic diagram of the modeled system is shown in Figure 7.3.3. The corresponding mass and heat balance of the systems are as follows28 Rmass, i = ρ i b C

dz i = k m Csat − Ci∞ i dt

(

)

[7.3.1]

k k   dT = h T − T ∞ − ∑ Rmass, i ( − δHevap ) ρ f b f CPf + ∑ ρ i z i bCCPi  i i =1 i =1   dt

(

)

[7.3.2]

where: Rmass,i zi ρi b Csat i C∞i km h T T∞ δH Cp

evaporation rate of component i volume fraction of component i density of pure component i thickness saturated solvent concentration of component i solvent concentration of component i in the bulk air mass transfer coefficient heat transfer coefficient temperature of coated film temperature of drying air latent heat of solvent heat capacity

Subscript f and C mean the substrate and coating layer respectively. Equation 7.3.1 and 7.3.2 apply to the each component of coating. 7.3.2.2 Liquid-vapor equilibrium The equilibrium saturated solvent concentration is related to the concentration of solvent at the coating surface by thermodynamic equilibrium relations, such as Henry’s law, Raoult’s law and the Flory-Huggins equation.8 The Raoult’s law is Csat = γizi i

Pisat RT

[7.3.3]

where: γi Pisat R

activity coefficient of component i saturated vapor pressure of component i gas constant

The saturated vapor pressure is calculated from the Antoine equation at the specific temperature. log10 P sat = A −

B T +C

[7.3.4]

390


The Antoine equation coefficients A, B, C for various organic solvents can be found elsewhere.19 The UNIFAC group contribution method was used to calculate the activity coefficient of each solvent.49 7.3.2.3 Heat and mass transfer coefficient The rate of heat transfer to the coating depends on the two factors as shown in Equation 7.3.2: the difference between the temperature of coating and ambient air (the driving force) and the geometry where the heat transfer occurs (heat transfer coefficient). The heat transfer coefficient is a function of the nozzle geometry and blowing air properties. Many variables affect the heat transfer coefficient of nozzles, such as nozzle geometry and size, nozzle to coating surface distance, nozzle to nozzle spacing, velocity of air at the nozzle exit and air motion above the coating surface.37,45 Therefore the average heat transfer coefficient of a zone can be expressed as follows, h = f (Geometry of nozzle, Properties of air ) × f (w)

[7.3.5]

where: h w

average heat transfer coefficient of a zone velocity of air at the nozzle exit.

Figure 7.3.3. Schematic diagram of modeled drying of coated film. [After reference 28].

The accuracy of drying rate calculation greatly depends on the proper estimation of the heat and mass transfer coefficient of nozzles. Many researches have been done to find out the heat transfer coefficient of nozzles, among them the Martin’s correlation is the notable one which correlates the geometry of impinging jet nozzle and air velocity to the heat and mass transfer coefficients.37 For a multiple slot jet nozzles, which is depicted in Figure 7.3.4, Martin suggested following empirical correlation,


391

Figure 7.3.4. Geometry of slot nozzles for the Martin’s correlation [After reference 37].

  Nu 2 3 / 4  4 Re = fO  f fO Pr 0. 42 6  + f  fO

     

2/3

[7.3.6]

where: Nu Re Pr f fO B LT H

average Nusselt number over a zone (2hB/κa) Reynolds number (2wB/νa) Prandtl number of air (νa/αa) Fraction open area (B/LT) [60 + 4(H/2B - 2)2]-1/2 Nozzle slot width (Figure 7.3.4) Nozzle spacing (Figure 7.3.4) Nozzle to coating surface distance (Figure 7.3.4)

Range of applicability is 1,500≤Re≤40,000 0.008≤f≤2.5fO 2≤H/B≤80 The heat transfer coefficient of the arrays of round jet nozzles and the other shapes of nozzles can be found elsewhere.37,45 In general the heat transfer coefficient can be written in the following form9,10 h = Kwn

[7.3.7]

where: K n

a constant that depends on the physical properties of the air and geometric properties of the dryer 0.6 ~ 0.8 (depending on the nozzle geometry)

One can easily calculate the actual heat transfer coefficient of a zone by running a heavy gauge web through a dryer and measuring temperature rising of web using non-contacting infrared thermometer in the early part of the dryer where the web is heating up.10 The procedures of measuring and accompanied calculation are illustrated in Table 7.3.1. The mass transfer coefficient is related to the heat transfer coefficient through the Chilton-Colburn analogy.16 Thus, Sh

 Sc  =  N u  Pr 

0. 42

= (Le)

0. 42

or

km =

h Le 0. 42 ka

[7.3.8]

392


where: Sh Nu Sc Pr Le km ka

average Sherwood number of a dryer zone average Nusselt number of a dryer zone Schmidt number of air Prandtl number of air Lewis number of air average mass transfer coefficient of a dryer zone heat conductivity of air

Table 7.3.1. Procedure for calculating h in dryer. [Adapted, by permission, from Cohen, E. D. and E. B. Gutoff, Modern Coating and Drying Technology, VCH Publishers, Inc., New York, 1992]

1 2 3

Set dryer air conditions and measure air inlet temperature Run heavy gauge polyester web through dryer Install two infrared thermometers relatively close together at initial part of drying zone Run web at varying line speeds and measure temperature rise in web

4

 200hx  Ta − Tx  = exp  −  ρC tU  Ta − To p   where: Cp h t Ta To Tx U x ρ

heat capacity, heat transfer coefficient thickness of web temperature of hot air initial temperature of web temperature of web at distance x line speed, distance between infrared thermometers density of web

The typical value of heat transfer coefficient of modern industrial dryer ranges from 50 to 150 J/m2secoC. Figure 7.3.5 shows the calculated heat transfer coefficient of an impingement dryer with varying nozzle exit velocity of air and fraction open area (nozzle spacing), calculation was done according to the Martin’s correlation. 7.3.2.4 Prediction of drying rate of coating The drying rate of coating and the subsequent residual solvent amount along with dryer length can be found by applying above equations. In magnetic media manufacturing pro-


393

cess, magnetic particulate dispersed solution is coated on the PET (polyethlylene terephthalate) film. The wet coating thickness is about 5 to 6 µm and the thickness of substrate is 14 to 15 µm. The line speed is normally 400 m/min to 1000 m/min. Usually several kinds of solvents are used for the coating solution, here we used three kinds of solvents - toluene, methyl ethyl ketone (MEK) and cyclohexanone (CYC). Equations 7.3.1 and 7.3.3 are applied to each solvent component. The temperatures of dryer zones were 50 to 130oC, and the air velocities at the nozzle exit were 10 to 20 m/sec. Figure 7.3.5. Calculated heat transfer coefficient of industrial dryer The average heat transfer coeffiwith varying air velocity. cient of a zone could be found by applying Martin’s correlation for the slot nozzles, however in this case we had the empirical coefficient for the Equation 7.3.7 which was supplied by the dryer nozzle manufacturer. And we obtained average heat transfer coefficients which were ranging from 80 to 140 J/m2soC according to the air velocities of the zones. The coupled and non-linear set of equations is solved by standard numerical method (such as Runge-Kutta-Gill method). The temperature and concentration of coating at time 0 was given as the initial concentration of coating. The concentration of coating at the next time step was calculated by solving Equation 7.3.1 with assuming that there was no significant temperature change of coating. This is a plausible assumption if the time step is small enough. Then the result of concentration of coating was substituted to the Equation 7.3.2 and the temperature of coating at that time step was calculated. The calculated concentration and temperature profile of coating are given as in the Figure 7.3.6. The concentration of solvent gradually decreases along with the dryer length. The slope of solvent concentration profile at zone 1 was nearly constant, but the slop was changed within the zone 1 though there were no changes in the drying conditions. It was due to the fact that the less Figure 7.3.6. Predicted residual solvent and temperature profile volatile solvent (cyclohexanone) along with the dryer length.

394


began to evaporate only after significant amount of more volatile solvents (MEK, toluene) were evaporated in the first part of zone 1. Therefore the slope was changed in zone 1. The temperature profile shows the effect of evaporative cooling in the first part of dryer. The temperature of coating at zone 1 was lower than that of coating at the entrance of dryer, but after significant amount of solvent was evaporated, the temperature of coating began to rise and eventually it approached to the temperature of zone. The temperature and concentration profile successfully explain the drying regimes - constant drying rate period (CDRP) and falling drying rate period (FDRP). 7.3.2.5 Drying regimes: constant drying rate period (CDRP) and falling drying rate period (FDRP) As we mentioned earlier in this chapter, the drying process can be divided into two or four distinct and easily identifiable processing regions: pre-dryer, constant drying rate period (CDRP), falling drying rate period (FDRP) and equilibration, or simply CDRP and FDRP.10 The pre-dryer and equilibrium regions denote the part of dryer which is located between the coating station and the dryer, and between the dryer exit and the take up roll, respectively. In the early stages of drying of polymeric solution coating, the Figure 7.3.7. Characteristic drying curve of polymeric solution surface of coating is fully wetted (methanol - ethyl acetate - acrylic resin solution). with solvent, thus the evaporation of solvent takes place as if there were no polymer or solute in it, as in the case of evaporation of pure solvent. The only factor that affects the drying rate of coating is the external mass transfer resistance such as the diffusion process of solvent vapor into the bulk air. The drying rate increases proportional to the temperature and velocity of blowing air. All the heat input is used to supply the latent heat of evaporation, therefore with constant heat input the temperature of coating remains constant as shown in Figure 7.3.7. Figure 7.3.8. Typical drying rate curve (methanol - ethyl acetate - Figures 7.3.7 and 7.3.8 represent typical drying curve of polymeric acrylic resin solution).


395

coating solution. In CDRP, as shown in the figures the slope of solvent content loss with time (drying rate) and the coating temperature are nearly constant. These drying rate curves were obtained by measuring weight changes of coating in an oven with blowing hot air above the coating samples - see Section 7.3.3.1. However, as the solvent content decreases the drying rate of coating also decreases gradually as shown in Figure 7.3.8. In this falling rate regime, mass transfer Figure 7.3.9. Concentration and temperature dependence of the bi- within the coating becomes the nary diffusion coefficient of a polystyrene-toluene solution according to the free volume theory of Vrentas and Duda. [After limiting factor, and with constant heat input the temperature of coatReferences 17]. ing rises and drying rate decreases. The solvent is transported to the exposed coating surface by diffusion and convection, where the solvent evaporates to the bulk air in a polymeric coating system. In most drying processes of thin film coatings the convection of solvent within the coating is negligible, thus the diffusion of solvent is the only method for solvent to reach the coating surface. The diffusivity of solvent in a polymeric solution falls dramatically when the solvent content is low as shown in Figure 7.3.9. Therefore the diffusion process of solvent within the coating controls the rate of drying in FDRP. To understand and improve the drying process during the FDRP, it is important to estimate diffusion coefficient of solvent within the coating. Vrentas and Duda predicted the binary diffusion coefficient by using free volume theory.17 Figure 7.3.9 shows the binary diffusion coefficient of toluene in the polystyrene-toluene solution that was found by applying free volume theory. The diffusion coefficient falls by a number of magnitudes in the low solvent concentration range, and it increases with increasing temperature. By the way the concentration dependency of diffusion coefficient at the low solvent concentration range declines at a high temperature. For example, at the temperature of 100oC the diffusion coefficient of toluene falls 5 orders of magnitude (about 10-5) from its highest value, but at the temperature of 180oC it falls only one order of magnitude. This shows us that it is highly required to eliminate residual solvent of coating at the low solvent concentration range the temperature of oven should be high enough. Therefore the temperatures of final zones keep high enough to achieve complete dryness in a multiple zone dryer. Normally the temperature of oven is restricted by the onset of deformation of substrate such as wrinkles and it is also restricted by the properties of coated material, such as melting point of binders. The parameters for the free volume theory of binary solution systems can be found in the literatures,17,53-55,60 and they have been effectively used in modeling drying process.1,2,6-8 But there are many difficulties in estimating diffusion coefficient for the ternary systems. Until now, almost all the empirical and theoretical correlations are restricted to the binary solution systems.

396


7.3.3 MEASUREMENT OF THE DRYING RATE OF COATED FILM The drying rate of coating is easily measured by simple experimental equipment. In a controlled air condition, the weight loss of coating due to the solvent evaporation is measured by electrical balance and filed at the PC. The weight loss with time is converted to drying rate of coating per unit area. The commercialized experiment equipment is commonly used to obtain the drying rate, such as thermo-gravimetric analyzer (TGA). Nowadays commercial TGA equipped with FT-IR or gas chromatography is available and readily used to obtain not only the overall drying rates of coatings but also the relative drying rate of each solvent in a multi-solvent system. However it is difficult to obtain such data at a high air velocity, because the air stream disturbs the balance and cause to oscillate the balance reading. Practically the available air velocity of this kind would be 1 m/sec or lower. Fourier-Transform-Infrared (FT-IR) spectrometer is used to obtain the drying rate at a higher air velocities.47,50 FT-IR is commonly used for the analysis of organic materials. Recently, FT-IR is applied to measure the drying rate of coating. This specially designed FT-IR with coating apparatus and air blowing system made it possible to measure solvent content without disturbances of airflow, and moreover it enabled us to find the content of each solvent with drying proceeded.47,50 This can be used to study selective evaporation of solvent and phase separation phenomena in a multi-solvent system.50 The air velocity of industrial dryer is up to 20 m/s or more and the temperature of oven is normally up to 200oC. A specially designed drying chamber was suggested to measure solvent concentration in these drying conditions.52,57 The chamber is equipped with flame ionization detector (FID) total hydrocarbon analyzer, and the oven exit gas which contains the evaporated solvent flows to the analyzer. The drying rate is calculated by multiplying the solvent concentration with exit gas flow rate. It provides access to a wider range of drying conditions that approximate industrial conditions. However, it is much more difficult to find out the actual drying rate of coating in an industrial dryer. In a continuous industrial dryer it is impossible to measure the drying rate by any of the above methods, because the coated substrate is running through the dryer at the speed of several hundreds meter per minute. To measure the drying rate of coating in such a condition, the dryer exit gas of each zone is analyzed by gas chromatography. Then the drying rate of each zone is calculated by multiplying solvent concentration with exit gas flow rate. From the drying rate of each zone, the evaporated amount of solvent is calculated. Thus the solvent concentration of coating is found at the point of each zone end.28,30 7.3.3.1 Thermogravimetric analysis The drying rate of a coating could be easily found by measuring coating weight loss during drying in a laboratory. The set-up of experimental equipment is relatively easy, and the commercial equipment can be readily available such as thermogravimetric analyzer (TGA).5,53 As drying proceeded, the weight of coating decreases due to the solvent evaporation. The amount of solvent loss with time is monitored by the balance. The schematic representation of the equipment is shown in Figure 7.3.10. The sample, such as coated films or a tray that contains the coating liquid, is mounted on the balance to be monitored. The temperature of coating is measured by non-contact infrared thermometer. A thin wire type thermocouple can be used to measure the temperature, and the thermocouple is attached to the coating or sample tray. The air is made up by conventional blower and is heated up to a certain temperature by electrical heater. The air is supplied to the coating surface through


397

Figure 7.3.10. Schematic diagram of drying experiment apparatus.

ducts. To make the laminar airflow over the balance, the length of ducts and sizes are determined. The experiment is usually conducted in a low air velocity so as the air not to disturb the balance. Honeycomb style mesh is often helpful to filter the air and make the airflow a laminar one. It enabled us to do the drying experiment at a higher velocity of air. The air velocity can be as high as 1 ~ 2m/sec. Actually the air velocity of industrial dryer is much higher than the experimental conditions and the directions of flow are normally perpendicular to the coating surface, not parallel with coating surface. Thus we can hardly expect to conduct a quantitative simulation from this experiment, but we can find the characteristic drying curve and mechanisms of drying of the given materials. The drying rate of coating is the weight of solvent loss per time divided by the area of evaporation. Rmass = −

W − Wt dW ≈ − t+∆t Adt A∆t

where: Rmass W t ∆t A

evaporation rate weight of sample at a specific time time time interval between measurements evaporation area

[7.3.9]

398


Figure 7.3.11. Drying rate curve of low volatile solution: N-methylpyrrolidone (NMP) - LiCoO2 - PVDF solution.

Figure 7.3.12. Comparison of drying rate between coating solution and pure solvent : N-methylpyrrolidone (NMP) - LiCoO2 - PVDF solution and pure NMP.

The example of measurement is shown in Figure 7.3.7. The remained solvent means the percent of remained solvent to the total solvent load, and the drying rate of coating is readily calculated using Equation 7.3.9. The drying rate at the specific time equals the slope of the drying curve of Figure 7.3.7, and it is depicted in Figure 7.3.8. The drying rate of coating shows constant and falling rate period with solvent content decreases. The temperature of coating is nearly constant at the beginning of drying where the slope of weight loss remains constant. As shown in Figure 7.3.7, most of the solvent is evaporated during the constant rate period. Figure 7.3.11 is the drying rate profile according to the solvent content at the various drying temperatures. The coating solution is consisted of LiCoO2 and poly(vinylidene fluoride) (PVDF) solution. Polymer is dissolved in N-methylpyrrolidone (NMP), and this solu-


399

tion is originally prepared for the anode cell of the lithium ion battery. The solvent content represents the percent of remained solvent to the weight of non-volatile component, (W-Wo)/Wo, where W and Wo represent the wet weight during drying and dry weight, respectively. NMP has very high boiling temperature (about 202oC) and low volatility, thus the drying of coating shows Figure 7.3.13. Schematic diagram of FT-IR spectrometer for the measurement of solvent content. [Adapted, by permission, from long period of constant rate. Figure 7.3.12 shows the comparison Saure, R. and V. Gnielinski, Drying Technol., 12, 1427 (1994)]. of drying rate between pure solvent and coating solution. The drying rate of coating solution is nearly equal to that of pure solvent during the CDRP. But the drying rate of coating gradually decreases as the solvent concentration falls, while the evaporation rate of pure solvent is constant throughout the evaporation. 7.3.3.2 Rapid scanning FT-IR spectrometer analysis Recently Fourier-Transform-Infrared (FT-IR) spectrometer is applied to drying studies to obtain the drying rate at a higher air velocities.47,50 FT-IR is a widely used analytical instru-

Figure 7.3.14. An example of FT-IR spectra of drying film. The methanol bands disappear slowly while the polymer bands remain constant. The ratio of the band height contains information on concentration. Methanol contents spectrum1 54 g/m2, spectrum2 19 g/m2, spectrum3 3.7 g/m2, spectrum4 1.5 g/m2, and spectrum5 0 g/m2, PVAc content 57 g/m2. [Adapted by permission, from Saure, R. and V. Gnielinski, Drying Technol., 12, 1427 (1994)].

400


ment to analyze organic materials. Development of powerful methods for the quantitative analysis made it possible to use the FT-IR technique for the drying studies. Moreover, with the aid of rapid scanning FT-IR the concentration of coating is measured with high resolution and sensitivity. The FT-IR spectrometer was modified for the drying experiment as shown in Figure 7.3.13.47 A typical spectra of a coated film, which contains methanol and polyvinyl acetate (PVAc), is shown in FigFigure 7.3.15. Comparison between gravimetric and FT-IR data. ure 7.3.14. The typical absorption [Adapted, by permission, from Saure, R. and V. Gnielinski, Drying bands of PVAc and methanol can Technol., 12, 1427 (1994)]. be identified from the literature and is shown in the Figure 7.3.14 for a reference.47 Basically the plot of FT-IR spectra and the ratio of band heights between the solvent and nonvolatile material give the information of concentration. The calculated methanol contents along with methanol bands are also seen in the Figure 7.3.14. The comparison between FT-IR and gravimetric data shows a good agreement as shown in Figure 7.3.15.47 The calibration method affects the qualitative analysis result, so care should be given in selecting the spectra to evaluate, baseline correction of the spectra and selection of the wavenumber ranges to evaluate.47 Besides the usefulness of FT-IR method to measure the solvent content at the high airflow ranges, it gives the concentration of each solvent in a multi-solvent system. Suzuki et al. applied the rapid scanning FT-IR technique to obtain the individual solvent concentration of binary solvent containing coating.50 The process path and phase diagram can be drawn from the FT-IR data. Figure 7.3.16 shows the phase diagram of MEK-toluene-polyvinylchloride and polyvinylacrylate copolymer (VGAH) system. The initial and final coating composition are given as an initial condition and a measured residual solvent content of a coating respectively. With the aid of FT-IR technique the drying process path between the two points can be found, and it enables us to investigate the phase separation phenomena during drying and the mechanism of structure formation of coating. Figure 7.3.16 shows how the coating of various initial solvent compositions is dried. The drying process path of toluene rich coating, sample 1, go through inside region of spinodal line in the early drying stages, but the drying process path of MEK rich coating do not go through the inside of binodal or spinodal region until the most of the solvent is evaporated. While the appearance of sample 3 and 4 kept clear after dried up, sample 1 whose drying process path was number 1, looked like a frosted glass due to the phase separation.50 Though the FT-IR technique is useful to study the various kinds of drying phenomena, the application of FT-IR spectrometer is restricted to a certain solvent system because the bands of spectrometer of each component of a solution must be separated. While the above


401

Figure 7.3.16. Phase diagram of MEK-toluene-VAGH system with the different initial solvent compositions. The concentration is measured by rapid scanning FT-IR method. [Adapted, by permission, Suzuki et al., Proceedings of 9th International Coating Science and Technology Symposium, Delaware, USA, May 17 - 20, 1998, pp. 21-24].

example of PVAc-methanol system is suitable for the FT-IR spectroscopy because of their separated bands, it is difficult to extend FT-IR technique to the other solvent systems. 7.3.3.3 High-airflow drying experiment using flame ionization detector (FID) total hydrocarbon analyzer The gravimetric method is limited to a certain air velocity level due to the oscillation of balance in the high airflow stream. However most of the industrial drying process accomplished by passing the coating under high air velocity jet of hot air. Thus to simulate the industrial drying conditions and according drying phenomena better, high airflow drying experiment setup (HADES) was suggested by Cairncross et al.52,57 Low air velocity results in low heat transfer coefficient, the heat transfer coefficient of conventional laboratory drying experiments ranges from 1 to 10 J/m2soC, while that of industrial dryer is ranges about 20 to 200 J/m2soC. The high heat and mass transfer at the evaporation surface may result in ‘trapping skinning’ because the surface evaporation is too high in compared with the rate of diffusion of solvent within the coating. The heat transfer coefficient of HADES is up to 26.4 J/m2soC which is equivalent with that of usual single-side impingement dryers,52 and it was reported that the HADES successfully simulates the skinning phenomena with the solution of PMMA-toluene.52 HADES measures solvent concentration of exhaust gas from the sample chamber where the solvent is evaporated from the coating. Then the rate of evaporation is equal to the

402


Figure 7.3.17. Schematic of HADES. HADES contains several sections. [Adapted, by permission, Vinjamur and Cairncross, Presented at the AIChE national meeting, Dallas, Texas, November 1, 1999].

Figure 7.3.18. Schematic of sample chamber of HADES. Dry nitrogen flows in from the port on the left side and the exhaust flows out through the port on the right to a FID. [Adapted, by permission, Winward and Cairncross, 9th International Coating Science and Technology Symposium, Delaware, USA, May 17 - 20, 1998, pp.343 - 346].

solvent vapor concentration times the gas flow rate, and the solvent loss is found by integrating the evaporation rate. The HADES is consisted of several sections as shown in Figure 7.3.17; gas handling system, a sample chamber and several process measurements.52,57 The nitrogen gas flows into the sample chamber as shown in Figure 7.3.18. The temperature and rate of gas flow are controlled and the gas temperatures before and after the coating sample tray are measured with thermocouples. The coating temperature is measured with thermocouple which is installed at the coating sample tray, and the solvent laden gas flows into the total hydrocarbon analyzer which is equipped with flame ionization detector (FID). The concentration of solvent at the exhaust gas is measured by total hydrocarbon analyzer, which is calibrated with known concentration of solvent vapor (via solvent bubblers). Figure 7.3.19 and Figure 7.3.20 show the examples of HADES running.52 Figure 7.3.19 shows the measured solvent concentration of the PVAC-toluene system, as shown in the figure the residual solvent decreases according to the rate of airflow. However, over the 36 cm/s of airflow rate the drying rate wasn’t changed, above this airflow rate, the residual


403

Figure 7.3.19. Measured solvent loss of PVAC-toluene system by HADES. [Adapted, by permission, Vinjamur and Cairncross, Presented at the AIChE national meeting, Dallas, Texas, November 1, 1999].

Figure 7.3.20. Measured solvent loss of PMMA-toluene system by HADES. [Adapted, by permission, Vinjamur and Cairncross, Presented at the AIChE national meeting, Dallas, Texas, November 1, 1999].

404


solvent wasn’t decreased by enhancement of airflow rate. From these results we can infer that the drying rate is controlled by internal mass transfer above this airflow rate. In a case of PMMA-toluene system, the drying rate at the airflow rate of 72 m/sec is lower than that of airflow rate of 36 m/sec (Figure 7.3.20). These are the contrary results to what we expected. It shows that this system exhibits ‘trapping skinning’ at the high airflow rate. Thus the residual solvent of final dried coating may be minimized at middle airflow as shown in the Figure 7.3.20. As shown in these HADES experiments, it is often the solution to lower the airflow rate or temperature of drying air when trapping skinning is suspected to occur. 7.3.3.4 MEASUREMENT OF DRYING RATE IN THE PRODUCTION SCALE DRYER Numerous drying rate data have been obtained in laboratory drying experiments using thermo-gravimetric analyzers as shown in the previous chapters. However it was difficult to know the actual drying rate or solvent concentration profiles in the pilot or production scale dryer from these experimental results.

Figure 7.3.21. Construction of material balance to calculate the solvent concentration of the coating at each zone end from the measured solvent concentration of exit gas. [After reference 30].


405

However, in a multiple zone dryer the solvent concentration of coating at the end of each zone could be found by analyzing the dryer exit gas with measured airflow. This experimental method provides concentration profile of each solvent with drying as well as the total drying rate of coating. It gives precise measurement value at the early stages of drying in which the concentration of solvent in the dryer exit gas is high, and the length of each zone is long enough to ensure negligible intermixing of air between adjacent zones. The solvent concentrations of the exit gas are measured by portable gas chromatography (GC), and the gas samples are taken at the exhausted air duct of each zone. From these concentration data and the airflow rate of each zone, the evaporation rate of solvent at each zone is calculated. (Figure 7.3.21) Evaporation rate of solvent at zone k, ERk [kg/min] = Solvent concentration, Ckout [kg/m3]×Airflow rate, Qkout [m3/min] [7.3.10] The example of solvent concentration measurement and the accompanying results of drying rate calculation are shown in Table 7.3.2., and the accompanying specification of dryer and the formulation of coating solution is given in Table 7.3.3 and 7.3.4. The solvent concentrations at the exhaust duct are measured by gas chromatography(MTI Analytical Instrument 200, Portable GC). The solvent concentrations were measured three times at the same point, and the deviations from the average value were less than ±5%. Table 7.3.2 Zone

Concentration, ppm

Evaporation rate, kg/h

MEK

Tol.

Cyc.

MEK

Tol.

Cyc.

Total

1

1655

740

142

47.6

27.2

5.6

80.4

2

1150

988

359

40.5

44.4

17.2

102.0

3

485

1243

1866

14.0

46.0

73.5

133.6

4

308

157

2223

1.5

1.0

15.2

17.7

5, 6

269

99

278

6.6

3.1

9.4

19.2

Table 7.3.3 Operating conditions

Dryer specification

Drying air temperature, oC Line speed, m/min

50 ~ 120 900

Total number of zones

6

Total dryer length, m

45

Dry coating thickness, µm

2.6

Dryer type

Air floating dryer

406


Table 7.3.4 Weight fraction, %

Initial coating amount, g/m2

Pigment and binder

35.5

6.23

Methyl ethyl ketone

21.5

3.77

Toluene

21.5

3.77

Cyclohexanone

21.5

3.77

Component

The solvent concentration of coating at the entrance of dryer is known with coating formulation and the solvent concentration of final dried coating could be measured offline, thus the percent evaporated solvent at each zone can be calculated. In this example of drying of magnetic coated film, the final dried coating contains less than 1000 ppm solvent, so we assumed that all the coated solvent was evaporated within the dryer. Finally, we estimated how much percent of solvent that was evaporated at each zone. The fractional amount of the evaporated solvent at zone, k, is Fk =

ER k n

∑ ER k

[7.3.11]

k =1

where: Fk ERk n

fractional evaporated solvent amount in zone k evaporation rate at zone k number of zone

From these fractional solvent evaporation data we estimate the solvent concentration in the coated film at the end of each zone. The solvent content per unit area decreases according to the fractional solvent evaporation data at each zone, for example, if the fractional evaporation rate of solvent is 50% in zone 1 and the initial solvent content is given as 10 g/m2, then the solvent content at the end of zone 1 should be 5 g/m2, thus we can find the solvent concentration at the end of each Figure 7.3.22. Measured coating composition changes with drying. zone. Moreover the measurements and calculations are applied to the each solvent in multiple solvent systems, therefore we can find the solvent content of each solvent at the end of zone. The resulting coating composition changes with drying are shown in Figure 7.3.22. The theoretical evaporation rate to dry a given solvent load at the specified line speed can be calculated as in Figure 7.3.21. Theoretically and experimentally found evaporation


407

rate should be equal in ideal case. We can check the measurement error by comparing measured and theoretical evaporation rate through the whole dryer. ERper is calculated by equation A in Figure 7.3.21. n

ER per − ∑ ER k k =1

ER per

=ε

[7.3.12]

In most cases the value of ε is about 0.05 ~ 0.10. ERper is the theoretical evaporation rate which assumes that all the solvent is evaporated only in the dryer. But in the real situation, some of the solvent is evaporated before the dryer, such as in the coating head and the pre-zone (between the coating head and the first dryer zone), and some of the solvent vapor leaks out of the dryer through the gaps where the substrate running in and out. Therefore measured ER was less than ERper by about 10%. This experimental method gives precise results at the early stages of drying where the concentration of solvent at the exit gas is high. And it is a unique method to measure the solvent concentration of coating in the production and pilot scale dryer. 7.3.4 MISCELLANEOUS 7.3.4.1 Drying of coated film with phase separation As a final process of coating process, drying plays an important role for the quality of products. The structure of coating is determined during the drying process. The structure formation of coating depends on the history of drying (drying process path) which represents the composition changes of coating during drying. The drying process path depends on the drying conditions such as temperature and velocity of hot air, residence time of coating in the dryer, humidity or solvent concentration of drying air, the initial composition of coating etc. Figure 7.3.23 shows the two different structures of coating according to the extremely different drying conditions (cellulose acetate solution of 10 wt% prepared in acetone 80 wt%

Figure 7.3.23. Cross-sectional SEM images of final coating microstructures prepared under slow and fast rates of external gas-phase mass transfer conditions. [Adapted, by permission, Mahendra et al., Proceedings of 9th International Coating Science and Technology Symposium, Delaware, USA, May 17 - 20, 1998, pp. 177 - 180].

408


and water 20 wt%).35 Sample A was dried under the room conditions of free convection mass transfer, and sample B was dried under impinging high velocity air which corresponded to intense forced convection. Though the initial coating compositions were same for the two samples, the coating structures were totally different with each other after drying. As shown in the figure, sample A had a porous structure at the bottom (substrate side) and dense structure at the top (air side) while sample B had a dense structure trough the coating layer.35 The structure of coating is determined by the competition of phase separation and solidification phenomena. The final structure is determined by the onset of solidification.35 The phase behavior of polymer solution is divided into three regions according to the thermodynamic stability; binodal (metastable region), spinodal (unstable region) and stable region. And the area of unstable and metastable region decreases with increasing temperature of solution.48,49 The phase separation can also be induced by the saturation of polymer in the solution. When a polymer is saturated in a good solvent, we can easily observe the precipitation of polymer by adding some non-solvent to the system. The phase separation region can be found by thermodynamic relations such as Flory-Huggins equation,35 and it is also be measured by adding non-solvent and observing the turbidity of solution (cloud point method). In these polymer-solvent-solvent system, the drying induced phase separation can be explained by process path and phase equilibrium line. During the drying of coating the phase separation occurs when the isothermal drying path intersects two-phase region as shown in Figure 7.3.24. Let’s say that the good solvent is more volatile than the non-solvent, then as the drying proceeded the fraction of non-solvent weight to the good solvent weight is steadily increased. Thus the drying process path intersects the phase equilibrium line as line ② in Figure 7.3.24. However, if the Figure 7.3.24. Conceptual representation of phase separation phegood solvent is less volatile than nomena during drying. the non-solvent, then the process path doesn’t intersect the phase equilibrium line until the significant amount of solvent is evaporated (line ①). In most clear coating system, the drying induced phase separation is unwanted phenomena. Therefore formulators are careful about the selection of solvent to avoid the defect. However in some applications, phase separation is needed to have desirable properties for the coating. In the coating of adhesive layer of hot stamping foil, the layer should have discontinuous porous structure as shown in Figure 7.3.25.29 Humidified air can be used to control the process path. In most polymer-solvent coating systems, the water acted as


409

non-solvent. But it is difficult to handle or store the water laden coating solution, since the state of solution is very unstable. Therefore to supply water to the coating the highly humidified air is sometimes used at the early stages of drying (in most cases the first and the second zone of dryer). The water was condensed at the coating surface and it acted as non-solFigure 7.3.25. Surface structure of coatings with different drying vent. By adjusting air humidity conditions (SEM ×500). Humidity of air [kg water/kg dry air] : A - various kind of coating structures 0.062, B - 0.036, Initial solid content : 10.0 wt%. can be obtained as in Figure 7.3.25. The phase separation phenomena in polymer-polymer-solvent system can be called as polymer incompatibility.41,42,56,58 The relative solubility of two polymer in the common solvent determines the surface structure of dried coating. Diverse kinds of surface structure can be obtained by adjusting drying rate, substrate surface properties, relative solubility of solvent.56,58 The shapes of surface structure can be simulated by Cell Dynamic System.41,42 7.3.4.2 Drying defects The maximum line speed of drying and operating conditions of dryer are often restricted by the onset of defects. The drying related defects could be classified according to the cause of defects - stress induced defects, surface tension driven defects, defects caused by air motion. 7.3.4.2.1 Internal stress induced defects Evaporation of solvent is necessarily accompanied with the shrinkage of coating in volume as illustrated in Figure 7.3.26. This shrinkage can be only allowed in the coating thickness direction, because the adhesion of coating to the substrate prevents coating from shrinking in the plane of coating. Thus the stress is develFigure 7.3.26. The origin of internal stress in a conventional poly- oped in the plane of coating. This mer solution coating. stress brings about the defects which is known as curling, cracking and peeling.10-15,21-23,34,46 Curling and cracking The internal stresses are build up with drying of wet coating. When the coating has enough mobility, the stresses that are developed during drying can be relieved by flow. After the coating solidified stresses will build up in the lateral dimensions. The solidification point equals the concentration at which the glass transition temperature of coating has risen

410


to the experimental temperature.13-15 These stresses bring about the defects such as cracking, curling and peeling. If the local strength is overwhelmed by the local stress, the response is cracking. It is often called as ‘mud cracking’, because it looks like a field of mud in a dry period. And if the local Figure 7.3.27. Internal stress related defects. stresses exceed the adhesion strength of coating to a substrate, the response is peeling or delamination as shown in Figure 7.3.27. At the edge of the coating the internal residual strength rises sharply, hence the stresses bring about the curling of coating at the edge if the adhesion strength of coating is sufficient to endure the peeling or delamination. The curling and peeling are also used to measure the internal stress of coating.12,14,15,22 With solvent evaporation, the coating becomes to be concentrated with polymers, and the coating becomes to have solid like nature (viscoelasticity). Thus the stresses depend not only on the strain rate, but also strain. Strain is deformation from the stress-free state.49 After the solidification point, the solvent evaporation continues, so the stresses persist. Croll analyzed the origin of residual internal stress during drying and correlated the internal stress with the coating properties.13-15 The residual internal stress, σ, for a coating is σ=

E φs − φr 1− ν 3

[7.3.13]

where: E ν φs φr

Young’s modulus Poisson’s ratio volume fraction of solvent at the point of solidification volume fraction of solvent in the dried coating

As shown in the Equation 7.3.13, the internal stress depends on the difference of solvent volume fraction before and after solidification point. Croll’s experiment confirms that the internal stress does not depend on the coating thickness until the coating thickness is so large that the net force on the interface exceeds adhesion. At this point stress is relieved by peeling.12-15 Adding plasticizer can often be helpful to reduce the internal residual stress of coating, because it makes coating more flexible to the later stages of drying.13-15 The residual internal stress depends on the solvent volume loss from the solidification point, we can shift the solidification point to the later stages of drying by using plasticizers. Adding plasticizers makes the coating more flexible, but it is not always the desirable property of final products. The residual internal stress is a result of combined action of stress and stress relaxation process. Therefore to give a sufficient relaxation time for the coating, we often dry a coated film in a mild operating conditions, e.g., dry at lower temperature and velocity of air. Peeling If the local stresses exceed the adhesive strength of coating to a substrate, then the coating is delaminated from the substrate. Peeling easily occurs when the coating is thick. With increasing thickness, one can find a critical thickness where the peeling occurs spontaneously. This critical thickness for the spontaneous peeling can be used as a method of measuring internal residual stress.12


411

Croll illustrated that interfacial work of adhesion could be found by peeling test.12 If the coating is peeled at a constant rate, the rate of change of total internal energy of coating is equal to zero. Thus the peeling strength is related to the interfacial work of adhesion as follows (Figure 7.3.28), Figure 7.3.28. 90° Peel test configuration [After reference 12].

F = γ − tC UR b

[7.3.14]

where: γ b F tC UR

interfacial work of adhesion sample width (Figure 7.3.28) measured force of peeling thickness of sample (Figure 7.3.28) recoverable strain energy per unit volume stored in the coating

In a case of spontaneous peeling, γ = t p UR

[7.3.15]

where: tp

critical thickness for the spontaneous peeling

Interfacial work of adhesion, γ, can be measured by spontaneous peeling test using above equation. Spontaneous peeling thickness, tp, can be directly measured, and the recoverable strain energy, UR, for an elastic material under a one-dimensional strain is, 1 UR = Eε i2 2

[7.3.16] where: E modulus of coating internal strain in coating εi

Figure 7.3.29. Graphical representation of the equations governing the 90° peel test. [Adapted, by permission, from Croll, S.G., J. Coating Technol., 52, 35 (1980)].

As shown in Figure 7.3.29 the interfacial work of adhesion and recoverable strain can also be found by measuring peeling strength at several different coating thickness. If the adhesive strength of the interface exceeds the cohesive strength of one of the coating component materials, there occurs cohesive failure - the coating layer loses the adhesive strength within the layer itself. At a small thickness, cohesive failure often occurs during the peeling test illustrated in Figure 7.3.28.

412


The spontaneous peeling thickness can be found by extrapolating the experimental data to the zero peeling strength. The validity of interfacial work of adhesion and recoverable strain which is obtained from this experiment can be tested by comparing this spontaneous peeling thickness with obtained from independent measuring of spontaneous peeling thickness. To enhance the adhesive strength, the surface properties of substrate are very important. The surface treatments are often used to enhance the adhesive strength between the coating and the substrate, as well as to enhance the surface energy of substrate - improving wettability of coating. If adhesive problem occurs, check out the surface treatment processes such as flame, plasma, or corona.21 Corona treatment does not persist permanently, therefore it should be done in-line, the corona is often applied directly before the coating station. Sometimes a thin, high surface energy adhesive or subbing layer is coated on the substrate to improve the adhesion of coating.21 As often the case, this layer is coated in-line during manufacturing of substrate - such as in-line coating of acrylic resin during bi-axial extension of PET film. If the adhesive failure occurred when one used these kind of treated substrate, one should check that the correct side was used where the subbing layer was coated and the status of coated subbing layer was perfect. Simple peel tests with adhesive tape can give a clue to the coating status of subbing layer. First laminating the adhesive tape with substrate where the coated material is applied, then pull off the tape rapidly. If there are problems with the subbing layer such as the partial un-coating of subbing layer, then the coating will be peeled off according to the un-coated pattern. By analyzing un-coated pattern, one can find the steps of process which have problems. If the substrate is suspected, change the substrate lots. 7.3.4.2.2 Surface tension driven defects Defects can arise during drying of a coated film by building non-uniform surface tension gradient over the coating. They include convection or Benard cells, fat edges or picture framing, etc.

Figure 7.3.30. Surface tension driven defect: Convection cells.

Convection cells Convection or Benard cells can arise when the coating is heated from the bottom of the coating, and the density or surface tension gradients are built across the coating thickness. These gradients lead to convection cells which look like regular close-packed hexagonal surface patterns (Figure 7.3.30).10,21 The evaporative cooling can also be the cause of the temperature gradient. This gradient arises the density and surface tension gradient, which brings about the internal flows within the coating to form the convection or Benard cells.


413

The convection cells, which come from the surface tension gradient, can arise when the Marangoni number (Ma) exceeds 80.10 The Marangoni number is Ma =

(dσ / dT )(dT / dy)h 2 µκ

[7.3.17]

where: σ µ κ T h dT/dy

surface tension viscosity heat conductivity temperature thickness of coating temperature gradient in the thickness direction

Convection cells can arise at a lower Marangoni number when the coating thickness is above 2 mm.21 Convection or Benard cells can be reduced or eliminated by adjusting operating conditions and formulation of coating solution. The possibility of having convection cells is reduced at the following conditions21 • Lower surface tension of coating liquid • Reduce the thickness of the wet coating • Increase the viscosity of coating liquid • Reduce the drying rate by adding low volatile liquid or reducing the temperature of drying air When the thickness of wet layer of coating is less than 1 mm, as is often the case with almost all coatings, convection cells are almost always due to the surface tension gradients.10,21 However, it is not desirable to lower the drying rate because it decreases the line speed of coating process. Frequently it is helpful to use surfactant to reduce the convection cells due to the surface tension gradients. Care should be given in selection of the surfactant and the amount of it not to deteriorate the final quality of products. The amount of surfactant should be minimized, excess surfactants can migrate to the coating surface and react with humidity at a coating station to form a haze coating surface. Moreover, the remaining excess surfactant, which is in the final products, can migrate to the coating surface while the products in use in a some environmental conditions to ruin the final quality of the products. Fat edges Fat edges can be built during drying of coating with non-uniform surface tension distribution at the edges of coating. Figure 7.3.31 illustrates the mechanisms of fat edge defects. The edges are usually thinner than the bulk of coating, thus as the evaporation proceeded the concentration of polymer is increased faster at edges than in the bulk of coating. Usually the surface tension of solid is much higher than that of polymer, the surface tension is higher at the edges than the bulk. The higher Figure 7.3.31. Formation of fat edge or picture framing. surface tension at the edges will

414


cause to flow to the edges, giving fat edges. The surfactants are often helpful to eliminate the fat edges. 7.3.4.2.3 Defects caused by air motion and others Dryer bands At the early stages of drying, where the coating has enough mobility to flow, the coating layer is apt to being disturbed by the motion of drying air.21 This defect is observed in the dryer which uses the arrays of round jet nozzles rather than the slot nozzles. Often it looks like a lane running in the machine direction and the width of bands is nearly equal to the nozzle diameter. When the air comes out of slots, bands are less likely if the flow across the slot is uniform.10,21,25 The bands are more likely to be developed with low viscosity solution, thick coating and high air velocity. Thus we can reduce the bands by reducing air velocity (but this also decreases the drying rate) and by increasing the initial solid content so as to increase the viscosity of solution and to coat thinner layer with concentrated solution. The geometry of arrays of round jet nozzles should be designed to avoid this defect by applying nozzles of larger diameters in the first part of zones. Large amount of fresh air is needed in the first part of zones of the multiple zone-dryer, because in the first zone the amount of evaporated solvent vapor is large, hence one should supply enough fresh air to ensure the dryer is operated below the lower explosive level (LEL) of a given solvent. Therefore the diameters of round jet nozzles are gradually decreased along the dryer length. Skinning If the drying rate of solvent is extremely high, then the solvent concentration at the surface of coating falls rapidly, for the rate of solvent evaporation is much higher than that of diffusion of solvent within the coating. This rapid decrease of solvent concentration at the surface causes to form a thin solid layer near the coating surface. This phenomena called as ‘skinning’, and it retards the drying of coating.4,5,8-10 Skinning can be reduced by using solvent laden gas as a drying air,4,8,10 but it isn’t applicable to the conventional drying process in which uses the volatile organic compound as a solvent. Most of the solvent has the possibility of explosion above a certain concentration in the air. Thus solvent laden drying gas should be applied in a inert environment, in which no oxygen is present. If the skinning occurs in a aqueous coating system, it is possible to use humid air to reduce the skinning. 7.3.4.3 Control of lower explosive level (LEL) in a multiple zone dryer If the solvent concentration of a zone exceeds certain level, then the system becomes to be in danger of explosion due to the flammability of organic solvent vapor. This level of solvent concentration is called as lower explosive level (LEL), and the dryer should be operated below the LEL prior to any constraints. Thus each zone needs sufficient airflow rate to meet these needs. However, too much airflow rate results in waste of energy to heat the excess air, and it increases the load of waste gas facilities (e.g., VOC emission control units). Therefore the airflow rate of each zone and the ratio of recycled to fresh air should be optimized. The typical airflow system for a multiple zone dryer is shown in Figure 7.3.32. The supplying air of each zone is consisted of fresh and recycled (used or returned from the exit gas) air as shown in the figure. LEL can be calculated if we know the evaporation rate of solvent of a zone. The evaporation rate of solvent can be measured or calculated as we explained in previous sections. Thus we can distribute the fresh air to each zone to meet the LEL safety without substantial increase in the total exhaust air.


Figure 7.3.32. Schematic diagram of airflow system to a zone. [After reference 30].

7.3.5 NOMENCLATURE B Nozzle slot width b Thickness [m] C Concentration of solvent at a gas phase [kg/m3] CP Specific heat [J/kg] E Young’s modulus ER Evaporation rate of solvent [kg/min] F measured force of peeling f Fraction open area (B/LT) fO [60 + 4(H/2B - 2)2]-1/2 G Mass flow rate of dry air [kg/sec] h Heat transfer coefficient [J/m2soC] H Nozzle to coating surface distance k Heat conductivity [J/moC] km Mass transfer coefficient [m/sec] LT Nozzle spacing P Vapor pressure [atm] Q Volumetric flow rate of air [m3/min] R Gas constant T Temperature [oC] t Time [s] tC Thickness of sample UR Recoverable strain energy per unit volume stored in the coating V Line speed [m/min] W Solvent coating amount [kg/m2] w Velocity of air at the nozzle exit [m/sec] WD Coating width [m] z Volume fraction of component [-] Dimensionless Numbers Sh Sherwood number

415

416


Nu Nusselt number Le Lewis number Sc Schmidt number Pr Prandtl number Ma Marangoni number Greek letters Volume fraction of solvent at the point of solidification φs Volume fraction of solvent in the dried coating φr Interfacial work of adhesion γ Activity coefficient γ Heat of vaporization [J/kg] δH Internal strain in coating εi Heat conductivity κ µ Viscosity Poisson’s ratio ν Density of pure component [kg/m3] ρ Surface tension σ Subscripts a Pertinent to dry air C Pertinent to coating layer f Pertinent to substrate, PET film fn Outlet condition i Component in Inlet condition Superscripts k Zone number Bulk air condition ∞

REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Alsoy, S. and J. L. Duda, Drying Technol., 16, 15, (1998). Aust, R., F. Dust and H. Raszillier, Chem. Eng. Process., 36, 469 (1997) Bird, R. B., W. E. Stewart and E. N. Lightfoot. Transport Phenomena, John Wiley & Sons, 1960. Bornside, D. E., C. W. Macosko, and L. E. Scriven, J. Appl. Phys., 66, 5185, (1989). Cairncross, R. A., A. Limbert, L. F. Francis, and L. E. Scriven, 24th annual meeting of Fine Particle Society, 24-28 August 1993, Chicago, IL. Cairncross, R. A., L. F.Francis, and L.E. Scriven, AlChE J., 42, 55 (1996). Cairncross, R. A., S. Jeyadev, R.F. Dunham, K. Evans, L.F. Francis, and L.E. Scriven, J. Appl. Polym. Sci., 58,1279 (1995). Cairncross, R. A., Ph.D. thesis, University of Minnesota, Minneapolis (1994). Cary, J. D. and E. B. Gutoff, “Analyze the Drying of Aqueous Coatings” Chem. Eng. Prog., 2, 73 (1991). Cohen, E. and E.Gutoff, Modern Coating and Drying Technology, VCH Publishers, Inc., New York, 1992. Cohen, E., E. J. Lightfoot and K. N. Christodoulou, Ind. Coating Res., 3, 45 (1995). Croll, S.G. J. Coating Technol., 52, 35 (1980). Croll, S.G. J. Coating Technol., 51, 64 (1979). Croll, S.G. J. Coating Technol., 50, 33 (1978). Croll, S.G. J. Appl. Polym. Sci., 23, 847 (1979). Cussler, E. L. Diffusion− Mass Transfer in Fluid Systems, 2nd Ed., Cambridge University Press, 1997. Duda, J. L., J. S. Vrentas, S. T. Ju, and H. T. Lu, AIChE J., 28, 279 (1982). Evans, K. J., Smith, W. R., Dunham, R. F., Leenhouts, T. J., Ceglinski; B. D., and Cairncross, R. A., US Patent 5,394,622. Felder, R. M. and R. W. Rousseau. Elementary Principles of Chemical Processes, 2nd Ed., John Wiley & Sons, Inc., 1986. Gutoff, E. B., Drying Technol., 14, 1673 (1996). Gutoff, E. and E. Cohen, Coating and Drying Defects - Troubleshooting Operating Problems, John Wiley & Sons, Inc., 1995.


22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

417

Hoffman, R.W., Surf. Interface Anal., 3, 62 (1981). Hu. S. M., J. Appl. Phys., 50, 4661 (1979). Huelsman, G. L. and W. B. Kolb, US Patent 5,694,701. Hunt, B. V., R. A. Yapel and R. K. Yonkoski, 9th International Coating Science and Technology Symposium, Delaware, USA, May 17 - 20, 1998, pp. 93 ~ 96. Imakoma, H. and M. Okazaki, Ind. Coating Res., 1, 101 (1991). Imakoma, H. and M. Okazaki, Ind. Coating Res., 2, 129 (1992). Kim, S. S. and M. H. Kwon, 9th International Coating Science and Technology Symposium, Delaware, USA, May 17 - 20, 1998, pp. 353 - 356. Kim, S. S., H. S. Park, I. C. Cheong and J. S. Hong, 3rd European Coating Symposium , Erlangen, Germany, Sep. 7 - 10, 1999. Kim, S. S., M. H. Kwon, submitted to the Drying Technol. (1999). Kolb, W.B. and G. L. Huelsman, 9th International Coating Science and Technology Symposium, Delaware, USA, May 17 - 20, 1998, pp.209 ~ 212. Kolb, W.B. and M.S. Carvalho, 9th International Coating Science and Technology Symposium, Delaware, USA, May 17 - 20, 1998, pp.9 ~ 12. Koschmieder, E.L., and M.I.Biggerstaff, J. Fluid Mech., 167, 49, (1986). Lei, H, L. F. Francis, W. W. Gerberich and L. E. Scriven, 9th International Coating Science and Technology Symposium, Delaware, USA, May 17 - 20, 1998, pp.97 ~ 100. Mahendra D., L. F. Francis and L. E. Scriven., 9th International Coating Science and Technology Symposium, Delaware, USA, May 17 - 20, 1998, pp 177 ~ 180. Mahendra D., L.F.Francis and L.E. Scriven., 1st Asian and 7th Japan Coating Symposium, Fukuoka, Japan, Sept. 1997, pp 8 ~ 9. Martin, H., Advances in Heat Transfer, 13, pp.1~60, Academic Press, New York, 1977. Ohta, T., H. Nozaki and M. Doi, J. Chem. Phys., 93, 2664 (1990). Oishi, T. and J. M. Prausnitz, Ind. Eng. Chem. Process Des. Dev., 17, 333 (1978). Okazaki, M., K. Shioda, K. Masuda, and R. Toei, J. Chem. Eng. Japan, 7, 99 (1974). Oono, Y. and S. Puri, Phys. Rev. A: At. Mol. Opt. Phys., 38, 434, (1988). Oono, Y. and S. Puri, Phys. Rev. A: At. Mol. Opt. Phys., 38, 1542, (1988). Pan, S. X, H. T. Davis, and L. E. Scriven, Tappi J., 78, 127 (1995). Pan, S. X., Ph.D. thesis, University of Minnesota, 1995. Polat, S. Drying Technol., 11, 1147 (1993). Sato, K., Prog. Org. Coat., 8, 143 (1980). Saure, R. and V. Gnielinski, Drying Technol., 12, 1427 (1994). Scriven, L.E. and W.J. Suszynski, Coating Process Fundamentals, Short course. June 19-21, 1996. University of Minnesota. Smith, J. M and H. C. Van Ness, Introduction to Chemical Engineering Thermodynamics, McGraw-Hill, 1987. Suzuki, I., Y. Yasui and A. Udagawa, 9th International Coating Science and Technology Symposium, Delaware, USA, May 17 - 20, 1998, pp. 21 - 24. Takase, K., H. Miura, H. Tamon and M. Okazaki, Drying Technol., 12, 1279 (1994). Vinjamur, M. and R. A. Cairncross, Presented at the AIChE national meeting, Dallas, Texas, November 1, 1999. Vrentas, J. S. and C. M. Vrentas, J. Polym. Sci. Part B : Polym. Phys., 32, 187 (1994). Vrentas, J. S. and J. L. Duda, AIChE J., 25, 1 (1979). Vrentas, J. S. and J. L. Duda, J. Appl. Polymer. Sci., 21, 1715 (1977). Walheim, S., Böltau, M., Mlynek, J., Krausch, G. and Steiner, U, Macromol., 30, 4995 (1997). Winward, T. and R. A. Cairncross, 9th International Coating Science and Technology Symposium, Delaware, USA, May 17 - 20, 1998, pp.343 - 346. Yamamura, M., K. Horiuchi, T. Kajiwara, and Adachi, K, 3rd European Coating Symposium, Erlangen, Germany, Sep. 7 - 10, 1999. Yapel, R. A., M.S. Thesis, University of Minnesota, Minneapolis, 1988. Zielinski J.M. and J. L. Duda, AIChE J., 38, 405 (1992).

8

Interactions in Solvents and Solutions Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli Dipartimento di Chimica e Chimica Industriale, Università di Pisa, Italy

8.1 SOLVENTS AND SOLUTIONS AS ASSEMBLIES OF INTERACTING MOLECULES A convenient starting point for the exposition of the topics collected in this chapter is given by a naïve representation of liquid systems. According to this representation a liquid at equilibrium is considered as a large assembly of molecules undergoing incessant collisions and exchanging energy among colliding partners and among internal degrees of freedom. The particles are disordered at large scale, but often there is a local order that fades away. Solutions may be enclosed within this representation. The collection of particles contains at least two types of molecules − those having a higher molar fraction are called the solvent − the others the solute. Collisions and exchange of energy proceed as in pure liquids, local ordering may be different, being actually dependent on the properties of the molecule on which attention is focused, and on those of the nearby molecules. Liquids at a boundary surface require more specifications to be enclosed in the representation. There is another phase to consider which can be either a solid, another liquid, or a gas. Further specifications must be added to characterize a specific boundary system (see Section 8.8), but here it is sufficient to stress that the dynamic collision picture and the occurrence of local ordering is acceptable even for the liquid portion of a boundary. This naïve description of liquid systems actually represents a model − the basic model to describe liquids at a local scale. As it has been here formulated, it is a classical model: use has been made of physical classical concepts, as energy, collisions (and, implicitly, classical moments), spatial ordering (i.e., distribution of elements in the space). It is clear that the model, as formulated here, is severely incomplete. Nothing has been said about another aspect that surely has a remarkable importance even at the level of naïve representations: molecules exert mutual interactions that strongly depend on their chemical composition. To say something more about molecular interactions, one has to pass to a quantum description.

420

Jacopo Tomasi, Benedetta Mennucci, Chiara Cappelli

Quantum mechanics (QM) is universally acknowledged as the appropriate theory to treat material systems at the level of phenomena of interest to chemistry. Therefore quantum mechanics is the legitimate theory level at which to treat liquids. We recall here the opening statement of a famous book on quantum chemistry:1 “In so far as quantum mechanics is correct, chemical questions are problems in applied mathematics.” Actually, the mathematics to apply to liquid systems is a hard nut to crack. Fortunately a sequence of many, but reasonable, approximations can be introduced. We shall consider and exploit them in the following section of this chapter. It is important to stress one point before beginning. The naïve picture we have summarized can be recovered without much difficulty in the quantum formulation. This will be a semiclassical model: the identity of the constituting particle is preserved; their motions, as well as those of nuclei within each molecule, are treated as in classical mechanics: the quantum methods add the necessary details to describe interactions. This semiclassical quantum description can be extended to treat problems going beyond the possibilities of purely classical models, as, for example, to describe chemical reactions and chemical equilibria in solution. The limits between the classical and quantum parts of the model are quite flexible, and one may shift them in favor of the quantum part of the model to treat some specific phenomenon, or in favor of the classical part, to make the description of larger classes of phenomena faster. We shall enter into more details later. It is sufficient here to underline that the choice of using a quantum approach as reference is not a caprice of theoreticians: it makes descriptions (and predictions) safer and, eventually, simpler. We shall start with the introduction of some basic simplifications in the quantum model.

8.2 BASIC SIMPLIFICATIONS OF THE QUANTUM MODEL The quantum mechanical description of a material system is obtained as solution of the pertinent Schrödinger equations. The first Schrödinger equation is the famous equation everybody knows: H(x)Ψ(x ) = EΨ(x )

[8.1]

The function Ψ(x) describes the “state of the system” (there are many states for each system). It explicitly depends on a set of variables, collectively called x, that are the coordinates of the particles constituting the system (electrons and nuclei). E is a number, obtained by solving the equation, which corresponds to the energy of the system in that state. H(x), called Hamiltonian, technically is an operator (i.e., a mathematical construct acting on the function placed at its right to give another function). Eq. [8.1] is an eigenvalue function: among the infinite number of possible functions depending on the variables x, only a few have the notable property of giving, when H(x) is applied to them, exactly the same function, multiplied by a number. To solve eq. [8.1] means to find such functions. The first step in the sequence of operations necessary to reach a description of properties of the system is to give an explicit formulation of the Hamiltonian. This is not a difficult task; often its formulation is immediate. The problems of “applied mathematics” are related to the solution of the equation, not to the formulation of H. The second Schrödinger equation adds more details. It reads:

8.2 Basic simplifications of quantum model

i£

∂ Ψ(x, t) = HΨ(x, t) ∂t

421

[8.2]

In practice, it expresses how the state Ψ(x) of the system evolves in time (i.e., it gives ∂Ψ(x)/ ∂t when Ψ(x) is known). Note that in eq. [8.1] the time t was not included among the parameters defining H (and Ψ): with eq. [8.1] we are looking at stationary states, not depending on time. In spite of this, eq. [8.1] may be applied to liquids, which are characterized by a continuous dynamic exchange of energy through collisions, and by a continuous displacement of the constituting molecules. This is not a problem for the use of the time independent formulation of the Schrödinger equation for liquids: eq. [8.1] actually includes kinetic energy and all related dynamic aspects. A direct use of eq. [8.1] for liquids means to treat liquids at equilibrium. We shall not make explicit use of the second equation, even when the nonequilibrium problem is considered; the use of the semiclassical approximation permits us to treat time dependent phenomena at the classical level, simpler to use. Let us come back to the expression of H(x). The number of parameters within (x) is exceedingly large if the system is a liquid. Fortunately, things may be simplified by using factorization techniques. When a system contains two subsystems (say A and B) that do not interact, the Hamiltonian can be partitioned so: H(x) = HA (x A ) + HB (x B )

[8.3]

This rigorously leads to a factorization of Ψ: Ψ(x) = ΨA (x A ) + ΨB (x B )

[8.4]

and to a partition of the energy: E =EA + EB

[8.5]

Equation [8.1] is in consequence transformed, without simplifications, into two simpler equations: HA (x A )ΨA (x A ) = E A ΨA (x A )

[8.6a]

HB (x B )ΨB (x B ) = E B ΨB (x B )

[8.6b]

In practice, there are always interactions between A and B. The Hamiltonian may always be written as: H(x) = HA (x A ) + HB (x B ) + HAB (x)

[8.7]

The relative magnitude of the coupling term permits us to distinguish among three notable cases. 1) If the coupling term is very small, it may be neglected, coming back to equations [8.6].

422


2)

If it is small but not negligible, it may be used to correct the solutions given by equations [8.6] with the use of appropriate mathematical tools: the factorized formulation with corrections continues to be simpler than the original one. 3) There are cases in which this factorization is profitably used, in spite of the fact that interactions between particles of A and B are of the same magnitude of interactions within both A and B. We shall now briefly introduce two factorizations of the last type that are of paramount importance in molecular quantum chemistry, then consider other factorizations of direct interest for the description of liquid systems. The first factorization we are introducing regards the electronic and nuclear coordinates (called r and R respectively) of the same system: H(x) = He(r) + Hn(R). It is clear that the electrostatic interaction between two electrons (and between two nuclei) within a given molecule is of the same magnitude as that concerning a couple electron-nucleus. In spite of this, the factorization is performed, with a tremendous effect on the evolution of quantum chemistry. Its physical justification rests on the very different masses nuclei and electrons have, and therefore on their velocities. It is often called Born-Oppenheimer (BO) approximation and in the following we shall use this acronym. The BO approximation makes possible the introduction of the concept of potential energy surface (PES), a very useful model to describe the motion of nuclei (often in the semiclassical approximation) taking into account the interactions with the quantum description of the electrons. Some more details can be useful in the following. To apply this factorization it is imperative to follow a given order. First, to solve the electronic equation with He(R) at a fixed geometry of nuclei: the output is a wave function and an energy both parametrically dependent on the nuclear geometry R given as input: Ψ e(r;R) and Ee(R) (actually there will be a set of electronic states, each with its energy). Second, to repeat the same calculations at different nuclear geometries R’, many times, until a sufficiently detailed description of the function Ee(R) is reached. Ee(R) is the PES we have mentioned: it may be used to define the potential operator within the nuclear Hamiltonian Hn(R) and then to compute vibrational and rotational states (another factorization is here introduced) or used in a semiclassical way to study the effect of nuclear motions. The second factorization we are introducing regards the electronic part of the system (after the BO factorization). The main procedure in use leads to factorization into many separate parts, each regarding one electron only: elec

H e (r ) = ∑ h i (ri ) i

Here again the coupling terms are of the same order of magnitude as the interactions left within each one-electron Hamiltonian (that explicitly regards the electrostatic interaction of the electron with all the nuclei of the systems, placed at fixed positions). The trick allowing this factorization consists in introducing within each one-electron Hamiltonian (the symmetry of electrons makes them all equal) an averaged description of the couplings, based on the yet unknown wave functions of the other electrons. The calculation proceeds iteratively: starting from a first guess of the averaged interaction, the description is progressively refined using intermediate values of the one-electron wave functions. The final out-

8.2 Basic simplifications of quantum model

423

put is a Ψe (r;R) expressed as an (anti)symmetrized product of the correct number of one-electron wave functions. Each electronic state will be described by its specific collection of such one-electron wave functions, called molecular orbitals (MO). The whole procedure is called SCF (self-consistent field), a name that reminds us of the technique in use, or HF (Hartree-Fock) to honor scientists working in its elaboration. This partition is the second cornerstone of molecular quantum chemistry. Almost all QM molecular calculations use the SCF theory (passing then, when necessary, to a higher level of the molecular QM theory). The concepts of MOs and of orbital energies derive from this one-electron factorization. In the following we shall make little explicit use of the two factorizations we have here examined, because we shall not enter into too technical details about how to obtain accurate molecular interaction potentials. They will be always in the background, however, and some concepts exposed here will be recalled when necessary. The last factorization we have anticipated in the introduction regards systems composed by many molecules, as molecular crystals, clusters and liquids. A factorization of a liquid into molecular subunits was implicit in the naïve model we have used in the introduction. The interactions within a molecule surely are larger than those among molecules; however, we cannot neglect these couplings, which are essential to describe a liquid. This is the main subject of this chapter and will be treated with due attention in the following sections. To complete this preliminary overview, we stress that the factorization techniques are quite flexible and that they may be applied at different levels. One among them deserves mention, because it will be used in the following. In the study of liquid systems there are several reasons to put more attention to a limited portion of the liquid, with the remainder of the liquid systems treated at a lower level of accuracy. We shall call the descriptions in which there is a portion of the whole system considered as the main component (called M) “focused models” while the remainder (called S) plays a supplementary or assisting role. In such an approximation, the Hamiltonian is partitioned in the following way: H(x) = HM (x M ) + HS (x S ) + HMS (x )

[8.8]

Focused models are used to study local properties in pure liquids, solutions, and interfaces. The largest use is to study solvation effects and reactions in solutions. In these cases M is composed of a solute supplemented by a solvation cluster (also a single solute molecule may be used). In these models the more detailed description of M is ensured by using the BO approximation, followed by a MO-based description of the electronic structure (there are also methods that replace the QM description of the electronic structure with some simpler semiclassical model). The S components are generally described with the aid of the intermolecular potentials we shall examine in the following sections. The description of the coupling takes into account the nature of the description chosen for both M and S. Focused models may also be used to get more detailed information on the structure of liquids, being in principle more accurate than descriptions solely based on intermolecular potentials. The computational cost is higher, of course, and this approach is now used only at the final stage of the assessment of models of the molecular interaction potential.

424


8.3 CLUSTER EXPANSION The pure ab initio approach we have summarized has never been used to determine the molecular interaction potentials. Further simplifications are generally used. We will start considering a model in which only the main component HM(xM) of the Hamiltonian [8.8] is considered. This means to pass from a liquid to a cluster of molecules. The presence of the molecules within M is explicitly acknowledged. Adding now the further constraint of keeping each molecule at a fixed geometry, the dimensionality of the R space is severely reduced. In fact, six coordinates for each molecule will be sufficient to define the relative position of its center of mass an its orientation with respect to an arbitrary fixed reference frame. If M is composed of n molecules, the BO coordinate space will have 3n-6 coordinates. It is convenient to consider now the whole PES EM(Rm), not limiting our attention to the minima (that correspond to equilibrium positions). For each point Rm of the hyperspace we introduce the following cluster expansion of the energy E M (Rm ) = ∑ E A + A

1 1 E ABC (RABC )+K [8.9] ∑ ∑ E AB (RAB ) + 6 ∑A ∑B ∑ 2 A B C

In this rather artificial (but exact) decomposition the whole energy is decomposed into the sum of the energies of the separate molecules, each at the geometry they have in the gas phase, followed by the sums of two-, three-, many-body terms. For each term of eq. [8.9] we have put in parentheses the indication of the pertinent nuclear coordinate subspace to emphasize that each dimeric interaction is defined in a 6-dimensional space (RAB), each trimeric interaction over a 9-dimensional space, etc., while EM is defined over a 3n-6 space. The total interaction energy is defined as the difference between [8.9] and the sum of the monomers’s energies: ∆E (12 , ,Km;Rm ) =

1 1 E AB (RAB ) + ∑ ∑ ∑ E ABC (RABC )+K [8.10] ∑ ∑ 2 A B 6 A B C

The convergence of this expansion is relatively fast for clusters composed by neutral molecules, less fast when there are charged species. In any case, it is not possible to interrupt the expansion to the two-body terms. This contribution gives the additive terms of the interaction energy; the other terms describe non-additive effects that in principle cannot be neglected. It is almost compulsory to proceed step by step and to study two-body interactions first. Each couple can be considered separately.

8.4 TWO-BODY INTERACTION ENERGY: THE DIMER The definition of a two-body potential is given by: ∆E AB (R) = E AB (R) − [E A + E B ]

[8.11]

We have kept the parameter R=RAB to underline that this potential depends both on the relative position of the two molecules and on their mutual orientation: it is a 6-dimensional function depending on these two sets of 3 parameters each, in the following indicated,

8.4 Two-body interaction energy

425

where necessary, with rab and Ωab. ∆EAB(R) has the status of a PES, with a shift in the reference energy, given here by the sum of the energies of the two monomers. A function defined in a 6-dimensional space is hard to visualize. Many devices have been introduced to render in graphic form selected aspects of this function. Some will be used in the following: here we shall use the simplest graphical rendering, consisting in fixing an orientation (Ωab) and two coordinates in the rab set in such a way Figure 8.1. Interaction energy for a dimer with respect to that the remaining coordinate corresponds the mutual approach distance at a fixed orientation. to the mutual approach between molecule A and B, along a given straight trajectory and a fixed mutual orientation. A typical example is reported in Figure 8.1. The energy curve may be roughly divided into three regions. Region I corresponds to large separation between interaction partners; the interaction is feeble and the curve is relatively flat. Region II corresponds to intermediate distances; the interactions are stronger compared to region I, and in the case shown in the figure, the energy (negative) reaches a minimum. This fact indicates that the interaction is binding the two molecules: we have here a dimer with stabilization energy given, in first approximation, by the minimum value of the curve. Passing at shorter distances we reach region III; here the interaction rapidly increases and there it gives origin to repulsion between the two partners. Before making more comments, a remark must be added. The shape of the interaction energy function in molecular systems is quite complex: by selecting another path of approach and/or another orientation, a completely different shape of the curve could be obtained (for example, a completely repulsive curve). This is quite easy to accept: an example will suffice. The curve of Figure 8.1 could correspond to the mutual approach of two water molecules, along a path leading to the formation of a hydrogen bond when their orientation is appropriate: by changing the orientation bringing the oxygen atoms pointing against each other, the same path will correspond to a continuously repulsive curve (see Figure 8.2 below for an even simpler example). We have to consider paths of different shape, all the paths actually, and within each path we have to consider all the three regions. To describe a liquid, we need to know weak long-range interactions as well as strong short-range repulsion at the same degree of accuracy as for the intermediate region. Studies limited to the stabilization energies of the dimers are of interest in other fields of chemical interest, such as the modeling of drugs. The ∆EAB(R) function numerically corresponds to a small fraction of the whole QM energy of the dimeric system. It would appear to be computationally safer to compute ∆E directly instead of obtaining it as a difference, as done in the formal definition [8.11]. This can be done, and indeed in some cases it is done, but experience teaches us that algorithms starting from the energies of the dimer (or of the cluster) and of the monomers are simpler and eventually more accurate. The approaches making use of this difference can be designed as variational approaches, the others directly aiming at ∆E are called perturbation approaches, because use is made of the QM perturbation theory. We shall pay more attention, here be-

426


low on the variational approach, then adding the basic elements of the perturbation theory approach. 8.4.1 DECOMPOSITION OF THE INTERACTION ENERGY OF A DIMER: VARIATIONAL APPROACH Coming back to Figure 8.1, its shape exhibiting a minimum makes it manifest that there are several types of interactions at work, with different signs and with different distance decays (at infinity all interactions go to zero). This subject has been amply studied and the main conclusions (there are details differing in the various schemes of interaction energy decomposition) are so widely known and intuitive that we feel authorized to introduce them here before giving a formal definition (that will be considered later). In Table 8.1 we report the names of these components of the interaction energy. Table 8.1. The main components of the bimolecular interactions ∆EAB(R) Component name Electrostatic

Acronym ES

Physical Meaning Coulomb interactions between rigid charge distributions

Induction

IND

Mutual electrostatic deformation of the two charge distributions

Exchange

EX

Quantum effect due to the Pauli exclusion principle

Dispersion

DIS

Interactions among fluctuations in the charge distributions of the two partners

Charge transfer

CT

Transfer of electrons between partners

All components of ∆EAB(R) are present in all points of the RAB 6-dimensional space we have introduced. It is convenient to examine them separately. We shall make references to a single decomposition scheme2 that we consider more convenient. Reference to other schemes will be done when necessary. Another view of methods may be found in Tomasi et al.3 The electrostatic term The ES term may be positive or negative: the shape of the ES(R) function strongly depends on the electric characteristics of the partners. If both molecules are reduced to dipoles, we have the two extreme situations (see Figure 8.2) which give rise to repulsive (i.e., positive) and attractive (i.e., negative) electrostatic contributions to ∆EAB, respectively. With different orientations of the two dipoles there will be different values of ES that in this simple case can be obtained with an analytical expression: ES(R, θ) = −2µ 1µ 2 cos θ / R 3

[8.12]

In more complex molecules, the ES(R) function has a complex shape that often determines the salient features of the whole ∆EAB(R) function. This is the reason why in the studies of molecular recognition and molecular docking, great attention is paid to a proper representation of ES(R). In the various methods for the decomposition of ∆EAB(R) there are no differences in the definition of ES. This quantity can be easily computed with ab initio methods, and the following recipe is the speediest and more used way of doing it.


427

Figure 8.2. ES for two dipoles with collinear (left) and antilinear (right) orientation.

The electronic wave functions of A and B are separately computed, each with its own Hamiltonian HA and HB. Ab initio methods give at every level of the formulation of the theory antisymmetric electronic wave functions, satisfying the Pauli exclusion principle, on which more details will be given later. The two antisymmetric wave functions ΨA and ΨB are then used, without modifications, in connection with the Hamiltonian of the whole system HAB to get the expectation value of the energy; according to the standard notation of quantum chemistry we can write: E I = ΨA ΨB HAB ΨA ΨB

[8.13]

It simply means the integral over the whole space of the complex conjugate of the function at the left (i.e., the function (ΨAΨB)* ) multiplied by the function HAB(ΨAΨB) (the application of an operator such as HAB to a function always gives a function). We can neglect complex conjugates (our functions are all real); the expression given above is a compact and clear indication of a set of operations ending with an integral, and we shall use it only to speed notations. HAB differs from the sum of HA and HB according to the following expression HAB = HA + HB + VAB

[8.14]

VAB collects terms describing electrostatic interactions between the nuclei and electrons of A with electrons and nuclei of B. The wave function ΨAB expressed as eigenfunction of HAB is of course antisymmetric with respect to all the electrons, of A as well as of B, but this has not yet been introduced in the model. The energy obtained with this recipe (the calculations are a by-product of the calculation of the dimer energy) may be so decomposed as: E I (R) = E A + E B + ES(R)

[8.15]

428


and so: ES(R) = E I (R) − (E A + E B )

[8.16]

The energies of the two monomers are already known, so the calculation of ES is immediate. Remark that in the right side of eq. [8.15] only ES depends on the geometry of the dimer, according to cluster model we are using. The induction term The second term of the interaction energy, IND, is always negative. IND is related to the mutual polarization of the electronic charge distributions of A and B (the nuclei are held fixed), inducing additional stabilizing effects. This induction (or polarization) energy contribution is defined in a different way in variational and perturbation theory approaches. Perturbation theory approaches are compelled to compute at the first order of the perturbation scheme only the effects due to the polarization of A with respect to the B distribution kept fixed, and in parallel the effects due to the polarization of B with A kept fixed. Mutual induction effects are introduced at higher order of the perturbation theory and have to be separated in some way from dispersion effects computed at the same time. In the variational approach use is made of an extension of the simple technique we have used for ES. The separation between electrons of A and B is maintained but the product ΨAΨB is now subjected to a constrained variational optimization using the Hamiltonian HAB. The two wave functions are so changed, allowing the effects of mutual polarization, because of the presence of the VAB term in the Hamiltonian: they will be so indicated as ΨAp and ΨBp . The resulting expectation value of the energy: E II (R) = ΨAp ΨBp HAB ΨAp ΨBp

[8.17]

may be so decomposed: E II (R) = E I (R) + IND(R)

[8.18]

and so IND(R) = EII(R) - EI(R). In this way, IND(R) contains all the mutual polarization effects. The exchange term The next term, EX, is positive for all the molecular systems of interest for liquids. The name makes reference to the exchange of electrons between A and B. This contribution to ∆E is sometimes called repulsion (REP) to emphasize the main effect this contribution describes. It is a true quantum mechanical effect, related to the antisymmetry of the electronic wave function of the dimer, or, if one prefers, to the Pauli exclusion principle. Actually these are two ways of expressing the same concept. Particles with a half integer value of the spin, like electrons, are subjected to the Pauli exclusion principle, which states that two particles of this type cannot be described by the same set of values of the characterizing parameters. Such particles are subjected to a special quantum version of the statistics, the Fermi-Dirac statistics, and they are called fermions. Identical fermions have to be described with an antisymmetric wave function; the opposite also holds: identical particles described by an


429

antisymmetric wave function are fermions and satisfy the Pauli exclusion principle. Introducing these concepts in the machinery of the quantum mechanical calculations, it turns out that at each coulomb interaction between two electrons described by MOs φ µ(1) and φυ (2) (the standard expression of this integral, see eq.[8.13], is: ) one has to add a second term, in which there is an exchange of the two electrons in the conjugate function: with a minus sign (the exchange in the label of the two electrons is a permutation of order two, bearing a sign minus in the antisymmetric case). There are other particles, called bosons, which satisfy other quantum statistics, the Bose-Einstein statistics, and that are described by wave functions symmetric with respect to the exchange, for which the Pauli principle is not valid. We may dispense with a further consideration of bosons in this chapter. It is clear that to consider exchange contributions to the interaction energy means to introduce the proper antisymmetrization among all the electrons of the dimer. Each monomer is independently antisymmetrized, so we only need to apply to the simple product wave functions an antisymmetrizer restricted to permutations regarding electrons of A and B at the same time: it will be called AAB. By applying this operator to ΨAp ΨBp without other changes and computing the expectation value, one obtains: EIII(R) = £AABΨAp ΨBp |HAB|AABΨAp ΨBp §

[8.19]

E III (R) = E II (R) + EX (R)

[8.20]

EX (R) = E III (R) − E II (R)

[8.21]

with

and

Morokuma has done a somewhat different definition of EX: it is widely used, being inserted into the popular Kitaura-Morokuma decomposition scheme.4 In the Morokuma definition EIII(R) is computed as in eq. [8.19] using the original Ψm monomer wave functions instead of the mutually polarized Ψmp ones. This means to lose, in the Morokuma definition, the coupling between polarization and antisymmetrization effects that have to be recovered later in the decomposition scheme. In addition, EX can be no longer computed as in eq. [8.20], but using EI energy: (i.e., EX' = EIII(R) - EI(R)). The problem of this coupling also appears in the perturbation theory schemes that are naturally inclined to use unperturbed monomeric wave functions, not including exchange of electrons between A and B: we shall come back to this subject considering the perturbation theory approach. The charge transfer term The charge transfer contribution CT may play an important role in some chemical processes. Intuitively, this term corresponds to the shift of some electronic charge from the occupied orbitals of a monomer to the empty orbitals of the other. In the variational decomposition schemes this effect can be separately computed by repeating the calculations on the dimer with deletion of some blocks in the Hamiltonian matrix of the system and tak-

430


ing then a difference of the energies (in other words: the same strategy adopted for the preceding terms, but changing the matrix blocks). In the Kitaura-Morokuma scheme CT also contains the couplings between induction and exchange effects. We consider unnecessary to summarize the technical details; they can be found in the source paper4 as well as in ref. [3] for the version we are resuming here. In standard perturbation theory (PT) methods, the CT term is not considered; there are now PT methods able to evaluate it but they are rarely used in the modeling of interaction potentials for liquids. The dispersion term The dispersion energy contribution DIS intuitively corresponds to electrostatic interactions involving instantaneous fluctuations in the electron charge distributions of the partners: these fluctuations cancel out on the average, but their contribution to the energy is different from zero and negative for all the cases of interest for liquids. The complete theory of these stabilizing forces (by tradition, the emphasis is put on forces and not on energies, but the two quantities are related) is rather complex and based on quantum electrodynamics concepts. There is no need of using it here. The concept of dispersion was introduced by London (1930),5 using by far simpler arguments based on the application of the perturbation theory, as will be shown in the following subsection. A different but related interpretation puts the emphasis on the correlation in the motions of electrons. It is worth spending some words on electron correlation. Interactions among electrons are governed by the Coulomb law: two electrons repel each other with an energy depending on the inverse of the mutual distance: e2/rij, where e is the charge of the electron. This means that there is correlation in the motion of electrons, each trying to be as distant as possible from the others. Using QM language, where the electron distribution is described in terms of probability functions, this means that when one electron is at position rk in the physical space, there will be a decrease in the probability of finding a second electron near rk, or in other words, its probability function presents a hole centered at rk. We have already considered similar concepts in discussing the Pauli exclusion principle and the antisymmetry of the electronic wave functions. Actually, the Pauli principle holds for particles bearing the same set of values for the characterizing quantum numbers, including spin. It says nothing about two electrons with different spin. This fact has important consequences on the structure of the Hartree-Fock (HF) description of electrons in a molecule or in a dimer. The HF wave function and the corresponding electron distribution function take into account the correlation of motions of electrons with the same spin (there is a description of a hole in the probability, called a Fermi hole), but do not correlate motions of electrons of different spin (there is no the second component of the electron probability hole, called a Coulomb hole). This remark is important because almost all the calculations thus far performed to get molecular interaction energies have been based on the HF procedure, which still remains the basic starting approach for all the ab initio calculations. The HF procedure gives the best definition of the molecular wave function in terms of a single antisymmetrized product of molecular orbitals (MO). To improve the HF description, one has to introduce in the calculations other antisymmetrized products obtained from the basic one by replacing one or more MOs with others (replacement of occupied MOs with virtual MOs). This is a proce-


431

dure we shall see in action also in the context of the perturbation theory. At the variational level considered here, the procedure is called Configuration Interaction (CI): each configuration corresponds to one of the antisymmetrized products of MO we have introduced, including the HF one, and the coefficients in front of each component of this linear expansion of the exact (in principle) wave function are determined by applying the variational principle. There are numerous alternative methods that introduce electron correlation in the molecular calculations at a more precise level that can be profitably used. We mention here the MC-SCF approach (the acronym means that this is a variant of HF (or SCF) procedure starting from the optimization no more of a single antisymmetric orbital product, but of many different products, or configurations), the Coupled-Cluster theory, etc., all methods based on a MO description of single-electron functions. For readers wishing to reach a better appreciation of papers regarding the formulation of interaction potentials, we add that there is another way of introducing electron correlation effects in the calculations. It is based on the density functional theory (DFT).6 There is a variety of DFT methods (detailed information can be found in the quoted monograph6); a family of these methods again makes use of MOs: they are called hybrid functional methods and give, on average, better results than other correlated methods at a lower computational cost. The introduction of electron correlation in the description of the monomers produces changes in their charge distribution and in their propensity to be polarized. For this reason ES, IND, and EX computed with correlated wave functions are somewhat different with respect to the values obtained with the corresponding HF wave functions. The procedure sketched with equations [8.13]-[8.21] can be adapted, with some modifications, to correlated wave function using a MO basis. The reader must be warned that here, as well as in other points of this chapter, we have simplified the discussion by omitting many details necessary for a proper handling and a fuller understanding the problem, but not essential to grasp the basic points. What is of more practical interest here is that HF descriptions of the dimer cannot give a DIS term. This may be recovered by introducing CI descriptions of the system. The simpler CI description, now largely used in routine calculations on molecules and molecular aggregates, is called MP2. This acronym means that use has been made of a specialized version of the perturbation theory (called Møller-Plesset) limited to second order to determine the expansion coefficients. MP2 wave functions contain elements able to give an appreciation of DIS, even if of limited accuracy. It is in fact possible to decompose MP2 values of ∆EAB(R) using the strategy we have outlined for the HF case, adding to each term the appropriate MP2 correction. Each term of the decomposition is somewhat modified, because of the MP2 corrections: the remainder of the MP2 contribution can be taken as a first approximation to DIS. There are other methods to get DIS values starting from HF wave functions, which may be more precise. Among them we quote the methods based on the response theory and on the use of dynamic polarizabilities. This powerful method has been developed during the years, and the outstanding contributions are due to McWeeny (1984) and Claverie (1986), to which reference is made for more details.7

432


The decomposition of the interaction energy through a variational approach: a summary We have so far given a description of the elements in which ∆EAB(R) may be decomposed using variational approaches. We report here the final expression: ∆E AB = ES + PL + EX + CT + COUP + DIS

[8.22]

At the HF level, all terms (but one: DIS) can be obtained with almost zero computational cost with respect to the numerical determination of ∆EAB; in the HF decomposition of the interaction energy there is a small additional term (called COUP) describing further couplings between the other components, which is often left as it is obtained (i.e., as the difference between the original ∆EAB value and the sum of the other four components) or subjected to further decompositions.8 The last term, DIS, may be calculated at this level using specialistic time-dependent formulations of the HF procedure. Going beyond the HF level, the decomposition can be obtained at the MP2 level essentially using the same techniques (DIS may be appreciated by separating some of the MP2 contributions that are strictly additive). The variational approaches we have considered are able to describe, and to decompose, the interaction energy at the level of accuracy one wishes, once the Figure 8.3. Decomposition of the interaction energy ∆E in necessary computational reH2OLHOH. sources are available. In fact, as we shall see later, there is no need of reaching extreme precision in the preliminary calculations to model interaction potentials for liquids, because other approximations must be introduced that will drastically reduce the accuracy of the description. We report in Figure 8.3 as an example the decomposition of the interaction energy of the water dimer, in the same orientation Ω used in Figure 8.1. At large distances the interaction is dominated by ES; this contribution also gives a reasonable approximation to ∆E at the equilibrium distance. The IND decays with the distance more rapidly than ES; this contribution is particularly sensitive to the quality of the expansion basis set {χ}. Old calculations of IND using restricted basis sets are not reliable. EX is a short ranged term: its contribution is however essential to fix the position of the minimum (and to describe portion III of the PES). CT and DIS terms have both a short-range character. This schematic analysis is valid for almost all the dimeric interactions where at least one partner has a dipolar character. The presence of a hydrogen bond (as is the case for the


433

example given in the figure) only introduces quantitative modifications, sufficient however to show by simple visual inspection if there is a hydrogen bond or not. Analogous trends in the decomposition of ∆E are present in the interactions involving charged-neutral species (also in the case of apolar molecules). A special case is given by the interaction of a molecule with the bare proton: in this case there is no EX contribution. If the partners have no permanent charge, or dipole, the interaction at large-medium distances is dominated by DIS, and by EX at small values of R. 8.4.2 BASIS SET SUPERPOSITION ERROR AND COUNTERPOISE CORRECTIONS Calculations of the interaction energies are affected by a formal error that may have important consequences on the final value of the energy and on its decomposition. We shall consider here the case of variational calculations for dimers, but the basic considerations can be extended to larger clusters and to other computational methods. The origin of the error is a non-perfect balance in the quality of the calculation of dimer energy, EAB and of energies of the two monomers, EA and EB. In fact, there are more computational degrees of freedom available for the dimer than for each monomer separately. The number of degrees of freedom corresponds to the number of basis functions available for the optimization of the electronic structure of the molecule, and hence for the minimization of the energy. Let us consider, to clarify the concept, the case of two water molecules giving origin to a dimer; each water molecule has ten electrons, while the quality and number of expansion functions is selected at the beginning of the calculation. This is called the expansion basis set (just basis set, or BS, for brevity) and it will be indicated for the molecule A with {χA}. The number of these basis functions is fixed, for example, 30 functions. The second molecule will be described by a similar basis set {χB} containing in this example expansion functions of the same quality and number as for molecule A (the two molecules are in this example of the same chemical nature). The wave function of the dimer AB and its energy will be determined in terms of the union of the two basis sets, namely {χAB}= {χA⊕ χB}, composed of 60 functions. It is evident that it is easier to describe 20 electrons with 60 parameters than 10 electrons with 30 only. The conclusion is that the dimer is better described than the two monomers, and so the dimer energy is relatively lower than the sum of the energies of the two monomers. This is called the basis set superposition (BSS) error. Why superposition error? When the two components of the dimer are at large distance, {χA} and {χB} are well separated, i.e., they have small superposition (or overlap), and so the relative error we are considering is modest, zero at infinity. When the two monomers are at shorter distances the superposition of the two basis sets increases (the basis functions are always centered on the pertinent nuclei) as well as the error. There is a simple recipe to correct this error: it consists of performing all the necessary calculations with the same basis set, the dimeric basis {χAB} which depends on the geometry of the dimer.9 This means that the energy of the monomers must be repeated for each position in the {R6} configuration space (see Section 8.4 for its definition). We add a superscript CP (counterpoise) to denote quantities modified in such a way and we also add the specification of the basis set. We replace eq. [8.11] with the following one:

[

]

CP CP ∆E CP AB (χ AB ; R) = E AB (χ AB ; R) − E A (χ AB ; R) + E B (χ AB ; R)

[8.23]

434


Figure 8.4. Location of the minimum energy for H2OLHF dimer according to the various basis sets and the various methods. Crosses (×) refer to SCF calculations without CP corrections, full circles (l) to CP corrected calculations. The ovoidal area refer to an estimate of the location of the minimum at the HF limit (including an estimate of the error). Left side: minimal basis sets; right side: double valence-shell basis sets.

∆ECP is by far more corresponding to the exact potential energy functions in calculations performed at the HF level with a basis set of a small-medium size. Passing to calculations with larger basis sets, the BSS error is obviously smaller, but the CP correction is always beneficial. We give in Figure 8.4 a graphical view of how CP corrections modify the equilibrium position of the dimer and, at the same time, its stabilization energy. The same holds for calculations performed at higher levels of the QM theory, beyond the HF approximation, with decreasing effects of the CP corrections, however. We may leave this last subject to the attention of the specialists because for the determination of interaction potential for liquids, the HF approximation is in general sufficient; in some cases it may be supplemented by applying simple levels of description for electron correlation, as we have already said. The CP corrected interaction energy may be decomposed into terms each having a definite physical meaning, in analogy with what we have exposed in the preceding subsection for ∆E without CP corrections. There are slightly different ways of doing it. We summarize here the strategy that more closely follows the physics of the problem.10 When this correction is applied, the various terms better satisfy chemical intuition in passing from one dimer to another of different chemical composition. Each term of the ∆E decomposition is corrected with an additive term ∆X (X stays for one of the components of the interaction energy) which is expressed as a difference in the monomers’ energies computed with the opportune basis set. Let us consider again the decomposition we have done in eq. [8.19].


435

The physical definition of ES (electrostatic interactions among rigid partners) clearly indicates that there is no room here for CP corrections, which would involve some shift of the monomer’s charge on the ghost basis functions of the partner. This physical observation corresponds to the structure of the blocks of the Hamiltonian matrix used to compute ES: no elements regarding the BS of the partner are present in conclusion and thus no corrections to ES are possible. In analogy, IND must be left unmodified. Physical analysis of the contribution and the structure of the blocks used for the calculation agree in suggesting it. CP corrections must be performed on the other elements of ∆E, but here again physical considerations and the formal structure of the block partition suggest using different CP corrections for each term. Let us introduce a partition into the BS space of each monomer and of the dimer. This partition can be introduced after the calculation of the wave function of the two monomers. At this point we know, for each monomer M (M stays for A or for B), how the complete monomer’s BS is partitioned into occupied and virtual orbitals φ M:

{χ Μ} = {φ0M

⊕ φvM

}

[8.24]

For the exchange term we proceed in the following way. The CP corrected term is expressed as the sum of the EX contribution determined as detailed above, plus a CP correction term called ∆EX: EX CP = EX + ∆EX

[8.25]

∆EX in turn is decomposed into two contributions: EX ∆EX = ∆EX Α + ∆B

[8.26]

with

[

(

EX ∆EX A = E A (χ A ) − E A χ A

)]

[8.27]

and a similar expression for the other partner. The CP correction to EX is so related to the calculation of another energy for the monomers, performed on a basis set containing occupied MOs of A as well as of B:

{χ } = {φ EX A

0 A

⊕ φ0B

}

[8.28]

We have used here as ghost basis the occupied orbitals of the second monomer, following the suggestions given by the physics of the interaction. The other components of the interaction energy are changed in a similar way. The ∆X corrections are all positive and computed with different extensions of the BS for the monomers, as detailed in Table 8.2.

436


Table 8.2 Explicit expressions of the ∆X counterpoise corrections to the interaction energy components

[ ( )] = [E (χ ) − E ( φ ⊕ φ )] = [E (χ ) − E ( φ ⊕ φ )] = [E (χ ) − E ( φ ⊕ φ )] = [E (χ ) − E ( χ ⊕ χ )] = [E (χ ) − E ( χ ⊕ χ )]

0 0 ∆EX A = E A (χ A ) − E A φA ⊕ φB

∆EX B ∆CT A ∆CT B

B

B

0 A

B

A

A

B

B

A

A

A

∆TOT B

B

B

0 A

B

0 B

B

∆TOT A

0 B

A

B

A

A

B

A

B

The correction ∆TOT permits to recover the full CP correction. It may be used to define correction: the ∆ COUP

(

∆COUP = ∆TOT − ∆EX + ∆CT

)

[8.29]

or the correction to the remainder, if COUP is further decomposed.8b This last step is of little utility for solvent-solvent interactions, but useful for stronger chemical interactions. There is no need of analyzing here these refinements of the method. The performances of the CP corrections to the interaction energy decomposition can be appreciated in a systematic study on representative hydrogen-bond dimers.11 A general review on the CP theory has been done by van Dujneveldt et al.12 (who prefer to use a different decomposition). 8.4.3 PERTURBATION THEORY APPROACH The perturbation theory (PT) approach aims at exploiting a consideration we have already expressed, namely, that ∆E is by far smaller than the sum of the energies of the separate monomers. It would be safer (and hopefully easier) to compute ∆E directly instead of getting it as difference between two large numbers. The formulation of the perturbation theory for this problem that we shall give here below has all the basic premises to satisfy this program. In practice, things are different: the introduction of other aspects in this formulation of PT, made necessary by the physics of the problem, and the examination and correction of finer details, put in evidence by the analysis of the results, make the PT approach more costly than variational calculations of comparable accuracy. Modern PT methods are competitive in accuracy with variational procedures but are rarely used at accurate levels to model potentials for liquids, the main reason being the computational cost. In spite of this we dedicate a subsection to PT methods, because this theoretical approach shows here, as in many other problems, its unique capability of giving an interpretation of the problem. To model interactions, we need, in fact, to have a clear vision of the physical elements giving origin to the interaction, and some information about the basic mathematical behavior of such elements (behavior at large distances, etc.). The formal perturbation theory is simple to summarize.


437

The theory addresses the problem of giving an approximate solution for a target system hard to solve directly, by exploiting the knowledge of a simpler (but similar) system. The target system is represented by its Hamiltonian H and by the corresponding wave function Ψ defined as solution of the corresponding Schrödinger equation (see eq. [8.1]) HΨ = EΨ

[8.30]

The simpler unperturbed system is described in terms of a similar equation H 0 Φ0 = E 0 Φ0

[8.31]

The following partition of H is then introduced H = H 0 + λV

[8.32]

as well as the following expansion of the unknown wave function and energy as powers of the parameter λ: Ψ = Φ 0 + λΦ (1) + λ2 Φ ( 2 ) + λ3 Φ ( 3 ) +K

[8.33]

E = E 0 + λE (1) + λ2 E ( 2 ) + λ3 E ( 3 ) +L

[8.34]

The corrections to the wave function and to the energy are obtained introducing the formal expressions [8.32]-[8.34] in the equation [8.30] and separating the terms according to their order in the power of λ. In this way one obtains a set of integro-differential equations to be separately solved. The first equation is merely the Schrödinger equation [8.31] of the simple system, supposed to be completely known. The others give, order by order, the corrections Φ (n) to the wave function, and E(n) to the energy. These equations may be solved by exploiting the other solutions, Φ1 , Φ 2 , Φ 3 ,K , ΦK , of the simpler problem [8.31], that constitute a complete basis set and are supposed to be completely known. With this approach every correction to Ψ is given as linear combination Φ ( n ) = ∑ CKn Φ k

[8.35]

K

The coefficients are immediately defined in terms of the integrals VLK = where the indexes L and K span the whole set of the eigenfunctions of the unperturbed system (including Φ0 where necessary). The corrections to the energy follow immediately, order by order. They only depend on the VLK integrals and on the energies E1, E2, ..., EK, ... of the simple (unperturbed) system. The problem is so reduced to a simple summation of elements, all derived from the simpler system with the addition of a matrix containing the VLK integrals. This formulation exactly corresponds to the original problem, provided that the expansions [8.33] and [8.34] of Ψ and E converge and that these expansions are computed until convergence. We are not interested here in examining other aspects of this theory, such as the convergence criteria, the definitions to introduce in the case of interrupted (and so approximate) expansions, or the problems of practical implementation of the method.

438


The MP2 wave functions we have introduced in a preceding subsection are just the application of this method to another problem, that of the electronic correlation. In this case, the simpler unperturbed system is the HF approximation, the corrections are limited to the second order, and the corresponding contributions to the energy are expressed as a simple summation of elements. The MP2 method is currently used in the PT description of the intermolecular potential:13 in such cases, two different applications of PT are used at the same time. We pass now to apply PT to the calculation of ∆EAB(R). The most reasonable choice consists of defining the unperturbed system as the sum of the two non-interacting monomers. We thus have: H 0 = HA + HB

[8.36]

Φ 0 = | Ψ0A Ψ0B >

[8.37]

E 0 = E 0A + E 0B

[8.38]

Φ0 is the simple product of the two monomers’ wave functions. The perturbation operator is the difference of the two Hamiltonians, that of the dimer and H0. The perturbation parameter λ may be set equal to 1: V = H − H 0 = VAB

[8.39]

The set Φ1 , Φ 2 , Φ 3 ,K , ΦK of the solutions of the unperturbed system can be obtained by replacing within each monomer wave function, one, two or more occupied MO with vacant MOs belonging to the same monomer. The perturbation operator V only contains oneand two-body interactions, and so, being the MOs orthonormal, the only VLK integrals different from zero are those in which L and K differ at the maximum by two MOs. The formulation is quite appealing: there is no need for repeated calculations, the decomposition of the interaction energy can be immediately obtained by separately collecting contributions corresponding to different ways of replacing occupied with virtual orbitals. Interrupting the expansion at the second order, one obtains the following result: ∆E ≈ E (1) + E (2) = ΨA ΨB |VΨA ΨB + −∑

| Ψ0A Ψ0B |VΨKA Ψ0B |2

−∑

| Ψ0A Ψ0B |VΨ0A ΨKB |2

E KA − E 0A

K

E KB − E 0B

K

− ∑∑ K

L

(E

I order: electrostatic term +

II order: polarization of A

+

II order: polarization of B

| Ψ0A Ψ0B |VΨKA ΨLB |2 A K

+ E LB ) − (E 0A + E 0B )

+

II order: dispersion A-B + higher order terms

[8.40]


439

We have here explicitly written the monomer wave functions, with a generic indication of their electronic definition: Ψ0A is the starting wave function of A, already indicated with ΨA, ΨKA is another configuration for A with one or two occupied one-electron orbitals replaced by virtual MOs. The first order contribution exactly corresponds to the definition we have done of ES with equation [8.16]: it is the Coulomb interaction between the charge distributions of A and B. The second term of the expansion, i.e., the first element of the second order contributions, corresponds to the polarization of A, due to the fixed unperturbed charge distribution of B; the next term gives the polarization of B, due to the fixed charge distribution of A. The two terms, summed together, approximate IND. We have already commented that perturbation theory in a standard formulation cannot give IND with a unique term: further refinements regarding mutual polarization effects have to be searched at higher order of the PT expansion. The last term of the second order contribution is interpreted as a dispersion energy contribution. The manifold of excited monomer states is limited here at single MO replacements within each monomer; the resulting energy contribution should correspond to a preliminary evaluation of DIS, with refinements coming from higher orders in the PT expansion. Remark that in PT methods, the final value of ∆E is not available. It is not possible here to get a numerical appraisal of correction to the values obtained at a low level of the expansion. All contributions are computed separately and added together to give ∆E. In conclusion, this PT formulation, which has different names, among which “standard” PT and RS (Rayleigh-Schrödinger) PT, gives us the same ES as in the variational methods, a uncompleted value of IND and a uncompleted appraisal of DIS: higher order PT contributions should refine both terms. One advantage with respect to the variational approach is evident: DIS appears in PT as one of the leading terms, while in variational treatments one has to do ad hoc additional calculations. Conversely, in the RS-PT formulation CT and EX terms are not present. The absence of CT contributions is quickly explained: RS-PT works on separated monomers and CT contributions should be described by replacements of an occupied MO of A with an empty MO of B (or by an occupied MO of B with an empty MO of A), and these electronic configurations do not belong to the set of state on which the theory is based. For many years the lack of CT terms has not been considered important. The attention focused on the examination of the interaction energy of very simple systems, such as two rare gas atoms, in which CT effects are in fact of very limited importance. The absence of EX terms, on the contrary, indicates a serious deficiency of the RS formulation, to which we have to pay more attention. The wave function ΦK used in the RS formulation does not fully reflect the electron permutational symmetry of the dimer: the permutations among electrons of A and B are neglected. This leads to severe inconsistencies and large errors when RS PT is applied over the whole range of distances. One has to rework the perturbation theory in the search of other approaches. The simplest way would simply replace Φ 0 = |ΨA0 ΨB0 > with |AABΦ 0 > where AAB is the additional antisymmetry operator we have already introduced. Unfortunately |AABΦ 0 > is not an eigenfunction of H0 as asked by the PT. Two ways of overcoming this difficulty are possible. One could abandon the natural partitioning [8.36] and define another

440


unperturbed Hamiltonian having |AAB Φ 0 > as eigenfunction. Changes in the definition of the unperturbed Hamiltonian are no rare events in the field of application of the perturbation theory and there are also other reasons suggesting a change in the definition of it in the study of dimeric interactions. This strategy has been explored with mixed success: some improvements are accompanied by the lack of well-defined meaning of the energy components when the expansion basis set grows toward completeness. A second way consists in keeping the original, and natural, definition of H0 and in changing the PT. This is the way currently used at present: the various versions of the theory can be collected under the acronym SAPT (symmetry adapted perturbation theory) It is worth remarking that in the first theoretical paper on the interaction between two atoms (or molecules) (Eisenschitz and London, 193014) use was made of a SAPT; the problem was forgotten for about 40 years (rich, however, in activities for the PT study of weak interactions using the standard RS formulation). A renewed intense activity started at the end of the sixties, which eventually led to a unifying view of the different ways in which SAPT may be formulated. The first complete monograph is due to Arrighini15 (1981), and now the theory can be found in many other monographs or review articles. We quote here our favorites: Claverie16 (1976), a monumental monograph not yet giving a complete formal elaboration but rich in suggestions; Jeziorski and Kolos17 (1982), short, clear and critical; Jeziorski et al.13 (1994), clear and centered on the most used versions of SAPT. SAPT theories are continuously refined and extended. Readers of this chapter surely are not interested to find here a synopsis of a very intricate subject that could be condensed into compact and elegant formulations, hard to decode, or expanded into long and complex sequences of formulas. This subject can be left to specialists, or to curious people, for which the above given references represent a good starting point. The essential points can be summarized as follows. The introduction of the intra-monomer antisymmetry can be done at different levels of the theory. The simplest formulation is just to use |AABΦ 0 > as unperturbed wave function, introducing a truncated expansion of the antisymmetry operator. This means to pass from rigorous to approximate formulations. One advantage is that the perturbation energies (see eq. [8.30]), at each order, may be written as the sum of the original RS value and of a second term related to the introduction of the exchange: (n ) (n ) + E exc E ( n ) = E RS

[8.41]

The results at the lower orders can be so summarized:

{

(1) (1) ∆E = E RS + E exc

} + {E

(2 ) RS

}

(2 ) + E exc +L

[8.42]

E (1) exc does not fully correspond to EX obtained with the variational approach. The main reason is that use has been made of an approximation of the antisymmetrizer: other contributions are shifted to higher order of the PT series. A second reason is that use has been here made of the original MOs ϕ i to be contrasted with the polarized ones ϕ pi , see Section 8.4.1: for this reason there will be in the next orders contributions mixing exchange and polarization effects. At the second order we have:

8.4 Two-body interaction energy (2 ) (2 ) (2 ) E exc = E exc + E exc − dis − ind

441

[8.43]

The partition of this second order contribution into two terms is based on the nature of the MO replacements occurring in each configuration ΨKm appearing in the sums of eq. [8.40]. These two terms give a mixing of exchange and dispersion (or induction) contributions. Passing to higher orders E(n) the formulas are more complex. In the RS part it is possi( 3) , but in general they are of mixed nature. The same happens ble to define pure terms, as ERSdis (n) for the E exc contributions. The examination of these high order contributions is addressed in the studies of the mathematical behavior of the separate components of the PT series. Little use has so far been made of them in the actual determination of molecular interaction potentials. 8.4.4 MODELING OF THE SEPARATE COMPONENTS OF ∆E The numerical output of variational decompositions of ∆E (supplemented by some PT decompositions) nowadays represents the main source of information to model molecular interaction potentials. In the past, this modeling was largely based on experimental data (supported by PT arguments), but the difficulty of adding new experimental data, combined with the difficulty of giving an interpretation and a decoupling of them in the cases of complex molecules, has shifted the emphasis to theoretically computed values. The recipes for the decomposition we have done in the preceding sections are too complex to be used to study liquid systems, where there is the need of repeating the calculation of ∆E(r,Ω) for a very large set of the six variables and for a large number of dimers. There is thus the need of extracting simpler mathematical expressions from the data on model systems. We shall examine separately the different contributions to the dimer interaction energy. As will be shown in the following pages, the basic elements for the modeling are to a good extent drawn from the PT approach. The electrostatic term ES may be written in the following form, completely equivalent to eq. [8.16] ES = ∫ ∫ ρ TA (r1 )

1 T ρ B (r2 )dr1dr2 r12

[8.44]

we have here introduced the total charge density function for the two separate monomers. The density function ρ TM ( r) is a one-electron function, which describes the distribution in the space of both electrons and nuclei. Formula [8.44] is symmetric both in A and B, as well as in r1 and r2. It is often convenient to decompose the double integration given in eq. [8.44] in the following way ES = ∫ ρ TA (r1 )VB (r1 )dr1

[8.45]

where VB (r1 ) = ∫ ρ TB (r2 )

1 r12

dr2

[8.46]

442


VB(r1) is the electrostatic potential of molecule B (often called MEP or MESP, according to the authors3,18,19). This a true molecular quantity, not depending on interactions, and it is often used to look at local details of the electrostatic interactions between molecules as required, for example, in chemical reactivity and molecular docking problems. To model simplified expressions of the intermolecular potential the direct use of eq. [8.45] does not introduce significant improvements. The MEP, however, may be used in two ways. We consider here the first, consisting of defining, and using, a multipolar expansion of it. The theory of multipolar expansion is reported in all the textbooks on electrostatics and on molecular interactions, as well as in many papers, using widely different formalisms. It can be used to separately expand ρA, ρB, and 1/r12 of eq. [8.44] or VB and ρA of eq. [8.45]. We adopt here the second choice. The multipolar expansion of VB(r1) may be so expressed: ∞

VB (r1 ) = ∑

l

∑M

l = 0 m= − l

B l, m

R − ( l+1)Y l m (θ, ϕ)

[8.47]

It is a Taylor expansion in powers of the distance R from the expansion center. There are negative powers of R only, because this expansion is conceived for points lying outside a sphere containing all the elements of the charge distribution. The other elements of [8.47] are the harmonic spherical functions Ylm and the multipole elements MBl,m which have values specific for the molecule:  4π  l m 3 m M lB, m =   ∫ r Y l (θ, ϕ)ρ B (r )d r = ΨB | M l ΨB  2l + 1

[8.48]

The spherical harmonics are quite appropriate to express the explicit orientational dependence of the interaction, but in the chemical practice it is customary to introduce a linear transformation of the complex spherical functions Ylm into real functions expressed over Cartesian coordinates, which are easier to visualize. In Table 8.3 we report the expressions of the multipole moments. Table 8.3. Multipole moments expressed with the aid of real Cartesian harmonics M0

1 element

Charge

Q

M

m 1

3 elements

Dipole

µx, µy, µz

M

m 2

5 elements

Quadrupole

θxy, θxz, θyz, θx 2 − y 2 , θz 2

M

m 3

7 elements

Octupole

ωxxy , etc.

The leading parameter is l, which defines the 2l poles. They are, in order, the monopole (l=0, a single element corresponding to the net molecular charge), the dipole (l=1, three elements, corresponding to the 3 components of this vector), the quadrupole (l=2, five components, corresponding to the 5 distinct elements of this first rank tensor), the octupole, etc. We have replaced the potential given by the diffuse and detailed charge distribution ρB with that of a point charge, plus a point dipole, plus a point quadrupole, etc., placed all at the expansion center. Note that the potential of the 2l pole is proportional to r-(l+1). This means that


443

the potential of the dipole decreases faster than that of the monopole, the quadrupole faster than that of the dipole, and so on. To get the electrostatic interaction energy, the multipolar expansion of the potential VB(r) is multiplied by a multipolar expansion of ρA(r). The result is: ∞

∞

ES ≈ ∑ ∑ Dll ′R − ( l+l ′+1)

[8.49]

l = 0 l ′= 0

where Dll ′ =

l

∑C

m= − l

m ll ′

A M lm M lB′m

[8.50]

is the interaction energy of the permanent dipole l of A with the permanent dipole l' of B (Cmll′ is a numerical coefficient depending only on l, l', and m). The calculation of ES via eq. [8.49] is much faster than through eq. [8.44]: the integrations are done once, to fix the M Xlm molecular multipole values and then used to define the whole ES(R) surface. For small size and almost spherical molecules, the convergence is fast and the expansion may be interrupted at a low order: it is thus possible to use experimental values of the net charge and of the dipole moment (better if supplemented by the quadrupole, if available) to get a reasonable description of ES at low computational cost and without QM calculations. We have, however, put a ≈symbol instead of = in eq. [8.49] to highlight a limitation of this expansion. To analyze this problem, it is convenient to go back to the MEP. Expansion [8.47] has the correct asymptotic behavior: when the number of terms of a truncated expression is kept fixed, the description improves at large distances from the expansion center. The expansion is also convergent at large values of R: this remark is not a pleonasm, because for multipole expansions of other terms of the interaction energy (as for example the dispersion and induction terms), the convergence is not ensured. Convergence and asymptoticity are not sufficient, because the expansion theorem holds (as we have already remarked) for points r lying outside a sphere containing all the elements of the charge distribution. At the QM level this condition is never fulfilled because eachρ(r) fades exponentially to zero when r → ∞. This is not a serious problem for the use of multipole expansions if the two molecules are far apart, and it simply reduces a little the quality of the results if almost spherical molecules are at close contact. More important are the expansion limitations when one, or both, molecules have a large and irregular shape. At strict contact a part of one molecule may be inserted into a crevice of the partner, within its nominal expansion sphere. For large molecules at close contact a systematic enlargement of the truncated expression may lead to use high value multipoles (apparent indication of slow convergence) with disastrously unphysical results (real demonstration of the lack of convergence). The introduction of correction terms to the multipolar expansions of VB or of ES, acting at short distances and called “penetration terms”, has been done for formal studies of this problem but it is not used in practical applications.

444


For larger molecules it is necessary to pass to many-center multipole expansions. The formalism is the same as above, but now the expansion regards a portion of the molecule, for which the radius of the encircling sphere is smaller than that of the whole molecule. Here it is no longer possible to use experimental multipoles. One has to pass to QM calculations supplemented by a suitable procedure of partition of the molecule into fragments (this operation is necessary to define the fragmental definition of multipoles with an analog of eq. [8.42]). There are many approaches to define such partitions of the molecular charge distribution and the ensuing multipole expansions: for a review see, e.g., Tomasi et al.20 It is important to remark here that such local expansions may have a charge term even if the molecule has no net charge (with these expansions the sum of the local charges must be equal to the total charge of the molecule). There is freedom in selecting the centers of these local expansions as well as their number. A formal solution to this problem is available for the molecular wave functions expressed in terms of Gaussian functions (as is the general rule). Each elementary electron distribution entering in the definition of Ψ, is described by a couple of basic functions χ *s χ t centered at positions rs and rt. This distribution may be replaced by a single Gaussian function centered at the well defined position of the overlap center: χu at ru. The new Gaussian function may be exactly decomposed into a finite local multipole expansion. It is possible to decompose the spherical function Y into a finite (but large) number of local multipole expansions each with a limited number of components. This method works (the penetration terms have been shown to be reasonably small) but the number of expansion centers is exceedingly large. Some expedient approximations may be devised, introducing a balance between the number of expansion centers and the level of truncation of each expansion. This work has been done for years on empirical bases. One strategy is to keep each expansion at the lowest possible order (i.e., local charges) and to optimize number and location of such charges. The modern use of this approach has been pioneered by Alagona et al.,21 using a number of sites larger than the number of atoms, with values of the charges selected by minimization of the difference of the potential they generate with respect to the MEP function VB(r) (eq. [8.40]). This is the second application of the MEP function we have mentioned. Alagona’s approach has been reformulated by Momany in a simpler way, by reducing the number of sites to that of the nuclei present in the molecule.22 This strategy has gained wide popularity: almost all the potentials in use for relative large molecules reduce the electrostatic contributions to Coulomb contributions between atomic charges. A relatively larger number of sites are in use for the intermolecular potentials of some simple molecules, as, for example, water, for which more accuracy is sought. Momany’s idea has led to a new definition of atomic charges. It would be possible to write volumes about atomic charges (AC), a concept that has no a precise definition in QM formalism, but is of extreme utility in practical applications. Many definitions of AC are based on manipulations of the molecular wave function, as, for example, the famous Mulliken charges.23 Other definitions are based on different analyses of the QM definition of the charge density, as for example Bader’s charges.24 There are also charges derived from other theoretical approaches, such as the electronegativity equalization, or from experimental values, such as from the vibrational polar tensors.


445

The charges obtained using Momamy’s idea of fitting the MEP with atomic charges are sometimes called PDAC (potential derived atomic charges). It has been realized that a fitting of MEP on the whole space was not convenient. It is better to reduce this fitting to the portion of space of close contact between molecules, i.e., near the van der Waals surface. This is the main technique now in use to define PDAC values. The induction term It is possible to define a molecular index PA for the induction term to be used in combination with the MEP VA to get a detailed description of the spatial propensity of the molecule to develop electrostatic interactions of classical type. Both functions are used under the form of an interaction with a unit point charge q placed at position r. In the case of VA this means a simple multiplication; in the case of PA there is the need of making additional calculations (to polarize the charge distribution of A). There are fast methods to do it, both at the variational level25a and at the PT level.25b The analysis of PA has not yet extensively been used to model IND contributions to ∆E, and it shall not be used here. This remark has been added to signal that when one needs to develop interactions potentials for molecular not yet studied interactions including, e.g., complex solutes, the use of this approach could be of considerable help. The multipole expansion of IND may be expressed in the following way: IND = IND(A ← B) + IND(B ← A)

[8.51]

with IND( A ← B ) ≈ −

1 ∞ ∞ − (l + l ′+ k + k ′+ 2 ) l < ∑ ∑R ∑ 2 ll ′=1 kk ′=1 m=− l
0.37, the emitting state changes to a much more polar, ICT state. The analysis of this spectroscopic data in the light of our scales reveals that, in fact, the solvatochromism of NR is normal, albeit dependent not only on the polarity of the solvent (SPP), but also on its acidity (SA) (see Figure 10.3.9): NR = -4.31(±0.57)SPP - 1.78(±0.19)SA + 21.72(±0.47) v max

with r = 0.974 and sd = 0.21 kK.

[10.3.26]

10.3 Solvent effects based on pure solvent scales

611

Figure 10.3.9. Plot of the fluorescence maxima, vmax(em), Figure 10.3.10. Plot of Grunwald and Wistein’s log for the Neutral Red in different solvents vs. the pre- ktBuCl values vs. the predicted log ktBuCl values according to eq. [10.3.29]. dicted, vmax(em), values according to eq. [10.3.26].

10.3.7.2.2 Kinetics Kinetics so closely related to the solvent effect as those of the Menschutkin reaction between triethylamine and ethyl iodide [eq. (10.3.27)], the solvolysis of tert-butyl chloride [eq. (10.3.28)] or the decarboxylation of 3-carboxybenzisoxazole [eq. (10.3.29)], are acceptably described by our scales.15,92,93 The equation for the Menschutkin kinetics is log ks/kHex = 8.84(±0.66)SPP + 1.90(±1.37)SA - 4.07(±0.53) [10.3.27] with n = 27, r = 0.947 and sd = 0.41. The rate of this reaction between triethylamine and ethyl iodide, which varies by five orders of magnitude from n-hexane (1.35×10-8 l mol-1 s-1) to DMSO (8.78×10-4 l mol-1 s-1), is accurately described by solvent polarity and acidity -the sensitivity to the latter is somewhat imprecise. The equation for the kinetics of solvolysis of tert-butyl chloride is: log k = 10.02(± 1.14)SPP + 1.84(± 0.99)SB + 8.03(± 0.69)SA - 19.85(± 0.70)

[10.3.28]

with n = 19, r = 0.985 and sd = 0.80. Based on eq. [10.3.28], all solvent effects increase the rate of solvolysis. However, the strongest contribution is that of polarity and the weakest one that of acidity. Although much less significant, the contribution of solvent basicity is especially interesting as it confirms that nucleophilicity also assists in the solvolytic process. Taking into account that it encompasses data spanning 18 orders of magnitude (from log k = -1.54 for water to log k = -19.3 for the gas phase), the fit is very good (see Figure 10.3.10). The equation for the kinetics of decarboxylation of 3-carboxybenzisoxazole is: log k = 10.37(± 1.47)SPP + 2.59(± 0.74)SB - 5.93(± 0.58)SA - 9.74(± 1.17)

with n = 24, r = 0.951 and sd = 0.73.

[10.3.29]

612

Javier Catalán

Equation [10.3.29] reproduces acceptably well the sensitivity of the decarboxylation rate of 3-carboxybenzisoxazole in pure solvents observed by Kemp and Paul.7 In addition, it clearly shows that such a rate increases dramatically with increasing polarity and, also, to a lesser degree, with solvent basicity. By contrast, it decreases markedly with increasing solvent acidity. This behavior is consistent with the accepted scheme for this decarboxylation reaction.93 10.3.7.2.3 Electrochemistry Gritzner94 examined the solvent effect on half-wave potentials and found those of K+ relative to bis(biphenyl)chromium(I)/(0), designated E1/2(BCr)K+, to be related to the Gutmann donor number (DN) for the solvents, so he concluded that K+ behaves as a Lewis acid against basic solvents. Based on the following equation, the behavior of E1/2(BCr)K+ is dictated largely by the basicity of the solvent but it is also dependent, however weakly, on its acidity: E1/2(BCr)K+ = -0.51(±0.05)SB + 0.12(±0.04)SA -1.06(±0.03)

[10.3.30]

with n = 17, r = 0.941 and sd = 0.03 V. It should be noted that solvent basicity increases the half-wave potentials of K+ whereas solvent acidity decreases it. 10.3.7.2.4 Thermodynamics In order to compare the shifts of the conformational equilibrium position with solvent effects it is advisable to select species exhibiting negligible cavity effects on the equilibrium position. Two firm candidates in this respect are the Scheme VI. conformational equilibria of 1,2,2-trichloroethane and the equilibrium between the equatorial and axial forms of 2-chlorocyclohexanone; both are accurately described by our scales.15 Of special interest is also the equilibrium between the keto and enol tautomers of pentane-2,4-dione (see the scheme VI). The ∆G values for this equilibrium in 21 solvents reported by Emsley and Freeman95 are accurately reproduced by solvent polarity and acidity according to ∆G = 18.45(±2.41)SPP + 4.22(±0.86)SA - 18.67(±1.96)

[10.3.31]

with n = 21, r = 0.924 and sd = 1.34 kJ mol-1. The equilibrium position is mainly dictated by solvent polarity; however, there is clearly a specific contribution of solvent acidity from the carbonyl groups of the dione. 10.3.7.3 Mixtures of solvents. Understanding the preferential solvation model The solvation of a solute, whether ionic or neutral, in a mixture of solvents is even more complex than in a pure solvent.1 This is assumed to be so largely because a solvent mixture


613

involves interactions not only between the solute and solvent but also among different molecules present in the mixture; the latter type of contribution also plays a central role in the solvation process. Among others, it results in significant deviations of the vapor pressure of a mixture with respect to the ideal behavior established by Raoult’s law. Solvation studies of solutes in mixed solvents have led to the conclusion that the above-mentioned divergences may arise from the fact that the proportion of solvent components may be significantly different around the solute and in the bulk solution. This would be the case if the solute were preferentially surrounded by one of the mixture components, which would lead to a more negative Gibbs energy of solvation.1,96 Consequently, the solvent shell around the solute would have a composition other than the macroscopic ratio. This phenomenon is known as “preferential solvation”, a term that indicates that the solute induces a change with respect to the bulk solvent in its environment; however, such a change takes place via either non-specific solute-solvent interactions called “dielectric enrichment” or specific solute-solvent association (e.g. hydrogen bonding). Preferential solvation has been studied in the light of various methods, most of which are based on conductance and transference measurements96, NMR measurements of the chemical shift of a nucleus in the solute97 or measurements of the solvatochromism of a solute in the IR98 or UV-Vis spectral region.99 Plots of the data obtained from such measurements against the composition of the bulk solvent (usually as a mole fraction) depart clearly from the ideal behavior and the deviation is ascribed to the presence of preferential solvation. Several reported methods aim to quantify preferential solvation;96-100 none, however, provides an acceptable characterization facilitating a clear understanding of the phenomenon. If preferential solvation is so strongly dictated by the polar or ionic character of the solute, then characterizing a mixture of solvents by using a molecular probe will be utterly impossible since any conclusions reached could only be extrapolated to solutes of identical nature as regards not only polarity and charge, but also molecular size and shape. However, the experimental evidence presented below allows one to conclude that this is not the case and that solvent mixtures can in fact be characterized in as simple and precise terms as can a pure a solvent. In the light of the previous reasoning, describing the solvolysis of tert-butyl chloride or the decarboxylation kinetics of 3-carboxybenzisoxazole in mixed solvents in terms of SPP, SB and SA for the mixtures appeared to be rather difficult owing to the differences between the processes concerned and the solvatochromism upon which the scales were constructed. However, the results are categorical as judged by the following facts: (a) The solvolysis rate of tert-butyl chloride in 27 pure solvents and 120 binary mixtures of water with methanol (31 mixtures), ethanol (31), isopropyl alcohol (1), trifluoroethanol (8), dioxane (13), acetone (27) and acetic acid (9), in addition to the datum for the gas phase, all conform to the following equation:92 log k = 10.62(±0.44)SPP + 1.71(±0.22)SB + 7.89(±0.17)SA - 20.07(±0.34) [10.3.32]

with n = 148, r = 0.99 and sd = 0.40. (b) The decarboxylation rate of 3-carboxybenzisoxazole in 24 pure solvents and 36 mixtures of DMSO with diglyme (4 mixtures), acetonitrile (4), benzene (7), dichloro-

614

Javier Catalán

methane (6), chloroform (6) and methanol (9) are accurately described by the following expression: log k = 10.03(±1.05)SPP + 2.41(±0.49)SB - 5.73(±0.40)SA - 9.58(±0.83) [10.3.33]

with n = 60, r = 0.990 and sd = 0.60. Other evidence obtained in our laboratory using solvent mixtures and probes as disparate in size and properties as the cation Na+ and Reichardt’s ET(30), also confirm that solvent mixtures are no more difficult to characterize than pure solvents. REFERENCES C. Reichardt, Solvents and Solvent Effects in Organic Chemistry, 2nd edn., VCH Publishers, Weinheim, 1988. 2 M. Berthelot, L. Péan de Saint-Gilles, Ann. Chim. Phys., 65, 385 (1862); 66, 5 (1862); 68, 225 (1863). 3 N.A. Menschutkin, Z. Phys. Chem., 5, 589 (1890). 4 E. Grunwald, S. Wistein, J. Am. Chem. Soc., 70, 846 (1948). 5 S. Wistein, A.H. Fainberg, J. Am. Chem. Soc., 79, 5937 (1957). 6 M.H. Abraham, R.M. Dogerty, M.J. Kamlet, J.M. Harris, R.W. Taft, J. Chem. Soc., Perkin Trans. 2, 913 (1987). 7 D.S. Kemp, K.G. Paul, J. Am. Chem. Soc., 97, 7305 (1975). 8 J.G. Kirkwood, J. Chem. Phys., 2, 351 (1934); 7, 911 (1939). 9 L. Onsager, J. Am. Chem. Soc., 58, 1486 (1936). 10 H. Block, S.M. Walker, Chem. Phys. Lett., 19, 363 (1973). 11 J.E. Brady, P.W. Carr, J. Phys. Chem., 89, 5759 (1985). 12 D.V. Matyushov, R. Schmid, B.M. Landansyi, J. Phys. Chem. B, 101, 1035 (1997). 13 C. Reichardt, Chem. Rev., 94, 2319 (1994). 14 R.S. Drago, J. Chem. Soc., Perkin Trans. 2, 1827 (1992). 15 J. Catalán, V. López, P. Pérez, R. Martin-Villamil, J.G. Rodriguez, Liebigs Ann., 241 (1995). 16 G. Taddei, E. Castelluci, F.D. Verderame, J. Chem. Phys., 53, 2407 (1970). 17. T.D. Epley, R.S. Drago, J. Am. Chem. Soc., 89, 5770 (1967); R.S. Drago, T.D. Epley, J. Am. Chem. Soc., 91, 2883 (1969); G.C. Vogel, R.S. Drago, J. Am. Chem. Soc., 92, 3924 (1970); R.S. Drago, L.B. Parr, C.S- Chamberlain, J. Am. Chem. Soc., 99, 3202 (1977); R.S. Drago, K.F. Purcell, Prog. Inorg. Chem., 6, 271 (1964); R.S. Drago, Coord. Chem. Rev., 33, 251 (1980); R.S. Drago, G.C. Vogel, J. Am. Chem. Soc., 114, 9527 (1992); R.S. Drago, Inorg. Chem., 32, 2473 (1993). 18 a) M.J. Kamlet, R.W. Taft, J. Am. Chem. Soc., 98, 377 (1976); b) R.W. Taft, M.J. Kamlet, J. Am. Chem. Soc., 98, 2886 (1976). 19 G.N. Lewis, Valence and Structure of Atoms and Molecules, The Chemical Catalog Co., 1923, p 142. 20 V. Gutmann, E. Vychera, Inorg. Nucl. Chem. Lett., 2, 257 (1966). 21 I.A. Koppel, V.A. Palm in Advances in Linear Free Energy Relationships (Eds. N.B. Chapman, J. Shorter) Chapter 5, Plenum Press, London, 1972, p 204. 22 E.M. Arnett, L. Joris, E. Michell, T.S.S.R. Murty, T.M. Gorrie, P. v R. Scheyer, J. Am. Chem. Soc., 96, 3875 (1974). 23 J. Catalán, J. Gómez, A. Couto, J. Laynez, J. Am. Chem. Soc., 112, 1678 (1990). 24 J. Catalán, P. Pérez, J. Elguero, W. Meutermans, Chem. Ber., 126, 2445 (1993). 25 E. Grunwald, S. Winstein, J. Am. Chem. Soc., 70, 846 (1948). 26 S. Winstein, A.H. Fainberg, J. Am. Chem. Soc., 79, 5937 (1957). 27 S. Winstein, E. Grunwald, H.W. Jones, J. Am. Chem. Soc., 73, 2700 (1951) 28 A.H. Fainberg, S. Winstein, J. Am. Chem. Soc., 78, 2770 (1956). 29 A.H. Fainberg, S. Winstein, J. Am. Chem. Soc., 79, 1597 (1957). 30 A.H. Fainberg, S. Winstein, J. Am. Chem. Soc., 79, 1602 (1957). 31 A.H. Fainberg, S. Winstein, J. Am. Chem. Soc., 79, 1608 (1957). 32 S. Winstein, A.H. Fainberg, E. Grunwald, J. Am. Chem. Soc., 79, 4146 (1957). 33 L.P. Hammett, J. Am. Chem. Soc., 59, 96(1937); and in Physical Organic Chemistry, 2nd Edition, McGraw-Hill, New York 1970. 34 A. Streitwieser Jr., Chem. Rev., 56, 617 (1956); Solvolytic Displacement Reactions, McGraw-Hill, New York 1962. 1


35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87

615

T.W. Bentley, P.v R. Schleyer, Adv. Phys. Org. Chem., 14, 32 (1977). T.W. Bentley, G. Llewellyn, Prog. Phys. Org. Chem., 17, 121 (1990). C.G. Swain, R.B. Mosely, D.E. Bown, J. Am. Chem. Soc., 77, 3731 (1955). T.W. Bentley, F.L. Schad, P.v R. Schleyer, J. Am. Chem. Soc., 94, 992 (1972). F.L. Schad, T.W. Bentley, P.v R. Schleyer, J. Am. Chem. Soc., 98, 7667 (1976). T.W. Bentley, G.E. Carter, J. Am. Chem. Soc., 104, 5741 (1982). I.A. Koppel, V.A. Palm in Advances in linear Free Energy Relationships, N.B. Chapman, J. Shorter Eds., Plenum Press, London, 1972, p 203 and references therein. G.A. Allerhand, P.V. Schleyer, J. Am. Chem. Soc., 85, 374 (1963). L.J. Bellamy, H.E. Hallam, Trans. Faraday Soc., 55, 220 (1959). L.J. Bellamy, H.E. Halam, R.L. Williams, Trans. Faraday Soc., 54, 1120 (1958). L.J. Bellamy, R.L. Williams, Proc. Roy. Soc.(London), A255, 22 (1960). C. Walling, P.J. Wagner, J. Am. Chem. Soc., 86, 3368 (1964). C. Somolinos, I. Rodriguez, M.I. Redondo, M.V. Garcia, J. Mol. Struct., 143, 301 (1986). K. Dimroth, C. Reichardt, T. Siepmann, F. Bohlmann, Liebigs Ann. Chem., 661, 1 (1963). M.A. Kessler, O.S. Wolfbeis, Chem. Phys. Lipids, 50, 51 (1989); C.J. Drummond, F. Grieser, T.W. Healy, Faraday Discuss. Chem. Soc., 81, 95 (1986). D.C. Dong , M.A. Winnick, Can. J. Chem., 62, 2560 (1984). K.W. Street, W E. Acree, Analyst, 111, 1197 (1986). I. Kristjánsson, J. Ulstrup, Chem. Scripta, 25, 49 (1985). E.M. Kosower, J. Am. Chem. Soc., 80, 3253 (1958). E.M. Kosower, J.A. Skorez, J. Am. Chem. Soc., 82, 2188 (1960). T.R. Griffiths, D.C. Pugh, Coord. Chem. Rev., 29, 129 (1979). Y. Marcus, Chem. Soc. Rev., 409 (1993). L.G.S. Brooker, G.H. Keyes, D.W. Heseltine, J. Am. Chem. Soc., 73, 5350-54 (1951). L.G.S. Brooker, A.C. Craig, D.W. Heseltine, P.W. Jenkins, L.L. Lincoln, J.Am. Chem. Soc., 87, 2443 (1965). J.E. Dubois, A. Bienvenue, J.Chim. Phys., 65, 1259 (1968). R.S. Drago, J. Chem. Soc., Perkin Trans II, 1827 (1992). R.S. Drago, M.S. Hirsch, D.C. Ferris, C.W. Chronister, J. Chem. Soc., Perkin Trans II, 219 (1994). R.S. Drago, Apllications of Electrostatic-Covalent Models in Chemistry, Surface Scientific Publishers, Gainesville, 1994. E. Lippert, Z. Elektrochem., 61, 962 (1957). J. Czekalla, W. Liptay, K.O. Meyer, Z. Elektrochem., 67, 465 (1963). V. Baliah, M.K. Pillay, Indian J. Chem., 9, 845 (1971). J. Catalán, V. López, P. Pérez, Liebigs Ann., 793 (1993). J. Catalán, J. Org. Chem., 60, 8315 (1995). J. Catalán. New. J. Chem., 19, 1233 (1995). J. Catalán, C. Díaz, V. López, P. Pérez, R.M. Claramunt, J. Phys..Chem., 100, 18392 (1996). J. Catalán, C. Díaz, Eur. J. Org. Chem., 885 (1999). J.S. Murray, S. Rauganathan, P. Politzer, J. Org. Chem., 56, 3734 (1991). J. Catalán, C. Díaz, V. López, P. Pérez, J.L.G. de Paz, J.G. Rodríguez, Liebigs Ann., 1785 (1996). J. Catalán, J. Palomar, C. Díaz, J.L.G. de Paz, J. Phys. Chem., 101, 5183 (1997). J. Catalán, C. Díaz, Liebigs Ann., 1942 (1997). J. Catalán, E. Mena, W. Meutermans, J. Elguero, J. Phys. Chem., 96, 3615 (1992). J.E. Kuder, D. Wychick, Chem. Phys. Lett., 24, 69 (1974). I. Guda, F. Bolduc, J. Org. Chem., 49, 3300 (1984). S.F. Mason, J.Chem. Soc., 1240 (1959). J.A. Paéz, N. Campillo, J. Elguero, Gazz. Chim. Ital., 126, 307 (1996). M.J. Kamlet, J.L.M. Aboud, R.W. Taft, J.Am. Chem. Soc., 99, 6027 (1977); ibid, 99, 8325 (1977). U. Mayer, V. Gutman, W. Gerger, Monasth. Chem., 106, 1235 (1975). V.I. Minkin, O.A. Osipov, Yu A. Zhadanov, Dipole Moments in Organic Chemistry, Plenum Press, New York, 1970. W.R. Fawcet, A.A. Kloss, J. Phys. Chem., 100, 2019 (1996). C.S. Giam, J.L. Lyle, J. Am. Chem. Soc., 95, 3235 (1973). J. Catalán, J.C. del Valle, C. Díaz, J. Palomar, J.L.G. de Paz, M. Kasha, Internat.J. Quam. Chem., 72, 421 (1999). A.F. Lagalante, R.J. Jacobson, T.J. Bruno, J. Org. Chem., 61, 6404 (1996). K.M.C. Davis, J. Chem. Soc. (B), 1128 (1967).

616

Victor Cherginets

F. Schneider, E. Lippert, Ber.Bunsen-Ges. Phys. Chem., 72, 1155 (1968). J. Catalán, V. López, P. Pérez, J. Fluorescence, 6, 15 (1996). J. Catalán, C. Díaz, V. López, P. Pérez, R.M. Claramunt, Eur. J. Org. Chem., 1697 (1998). M.K. Singh, H. Pal, A.C. Bhasikuttan , A.V. Sapre, Photochem & Photobiol., 68, 32 (1998). J. Catalán, C. Díaz, F. Garcia-Blanco, J. Org. Chem., 64, 6512 (1999). J. Catalán, C. Díaz, F. Garcia-Blanco, J. Org. Chem., in press. G. Gritzner, J. Phys. Chem., 90, 5478 (1986). J. Emsley, N.J. Freeman, J. Mol. Struc., 161, 193 (1987). H. Scheneider, in Solute-Solvent Interactions, J.F. Coetzee and C.D. Ritchie (Eds), Dekker, New York, Vol 1 p 301 (1969). 97 L. S. Frankel, C.H. Langford, T.R. Stengle, J. Phys. Chem., 74, 1376 (1970). J.F. Hinton, E.S. Amis, Chem. Rev:, 67, 367 (1967). 98 A.I. Popov, in Soluto-Solvent Interactions, J.F. Coetzee and C.D.Ritchie(eds) Dekker, New York, 1976, Vol 2 pg 271. 99 K. Dimroth , C. Reichardt, Z. Anal. Chem., 215, 344 (1966) J. G. Dawber, J.Ward, R. A. Williams, J. Chem. Soc., Faraday Trans. 1, 84, 713 (1988). 100 J. Midwinter, P. Suppan, Spectrochim. Acta, 25A, 953 (1969); P. Suppan, J. Chem. Soc., Faraday Trans.,1, 83, 495 (1987); M. W. Muanda, J.B. Nagy, O.B. Nagy, Tetrahedron Lett., 38, 3424 (1974); O.B. Nagy, M.W. Muanda, J.B. Nagy, J. Phys. Chem., 83, 1961 (1979): H. Strehlow, H. Schneider, Pure Appl. Chem., 25, 327 (1971); M.S. Greenberg, A.I. Popov, Spectrochim. Acta, 31A, 697 (1975); H. Langhals, Angew. Chem. Int. Ed. Engl., 21, 724 (1982); A. Ben-Nain, J. Phys. Chem., 93, 3809 (1989); P. Chatterjee, S. Bagchi, J. Chem. Soc., Faraday Trans, 87, 587 (1991); W.E. Acree, Jr., S. A Tucker, D. C. Wilkins, J. Phys. Chem., 97, 11199 (1993); R.D. Skwierczynski, K.A. Connors, J. Chem. Soc. Perkin Trans 2, 467(1994) M. Rosés, C. Ráfols, J. Ortega, E. Bosch, J. Chem. Soc. Perkin Trans 2, 1607 (1995); W. E. Acree Jr., J. R. Powell, S. A. Tucker, J. Chem. Soc. Perkin Trans 2, 529 (1995). 88 89 90 91 92 93 94 95 96

10.4 ACID-BASE EQUILIBRIA IN IONIC SOLVENTS (IONIC MELTS) Victor Cherginets Institute for Single Crystals, Kharkov, Ukraine

Ionic melts are widely used in the science and engineering as media for performing different processes such as electrolysis, electrochemical synthesis, single crystals growing, etc. Practically complete dissociation of ionic media to the constituent ions creates high current densities at electrolysis. The absence of oxidants, similar to H+, makes it possible to obtain products, which cannot be obtained from aqueous solvents (i.e., alkaline and alkaline earth metals, sub-ions, etc.). From the ecological standpoint, molten ionic media are especially available as technological solvents since their employment does not cause the accumulation of liquid wastes because cooling to the room temperature transforms ionic liquids into a solid state. Processes taking place in ionic melt-solvents are considerably affected by impurities contained in the initial components of the melt or formed during preparation (mainly, melting) of solvents due to the high-temperature hydrolysis of melts or their interactions with container materials (Al2O3, SiO2, etc.) or active components of atmosphere (O2, CO2, etc.). The list of these impurities is wide enough and includes multivalent cations of transition metals, different complex anions (oxo- or halide anions). The effect of the mentioned admixtures on the processes in ionic melts depends mainly on the degree of their donor-acceptor interactions with constituent parts of the melt.

10.4 Acid-base equilibria in ionic solvents

617

10.4.1 ACID-BASE DEFINITIONS USED FOR THE DESCRIPTION OF DONOR-ACCEPTOR INTERACTIONS IN IONIC MEDIA Donor-acceptor interactions in ionic media are often described as acid-base interactions according to Lewis1 and Lux-Flood2-5 definitions. The classic variant of the former definition considers acids as acceptors of electron pairs and bases as their donors. In modern variant of this definition, acids are the electron pair (or anion) acceptors or cation (proton) donors, bases are the electron pair (anion) donors or cation (proton) acceptors. Lux-Flood definition considers bases as donors of oxide-ion, O2-, its acceptors are Lux acids. 10.4.1.1 The Lewis definition The Lewis acid-base process may be described by the following scheme: A + :B = A:B, K

[10.4.1]

where: A B K

an acid, a base, the equilibrium constant (used below pK ≡ -log K).

This definition can be used for the description of interactions in ionic melts, containing complex anions undergoing heterolytic dissociation. Processes reverse to [10.4.1], i.e., the acid-base dissociation of A:B adducts (ions) in some solvents are considered as acid-base equilibria of these solvents. Alkaline chloroaluminate melts containing the excess of AlCl3 vs. the stoichiometry 3+ 6,7 + the excess of are promising solvents for preparing sub-ions, e.g., Cd+(Cd 2+ 2 ), Bi , Bi 5 , 8 acid (AlCl3) favors their formation. Tremillon and Letisse, Torsi and Mamantov9,10 studied the acid-base properties of molten mixtures AlCl3-MCl (M=Li, Na, K, Cs) with AlCl3 concentration exceeding 50 mol% in the temperature range 175-400oC. Equilibrium molarities of Cl- were determined by a potentiometric method with the use of a chloride-reversible electrode. The solvents undergo the acid-base dissociation according to the following equation:

(

)

2AlCl 4− = Al 2 Cl 7− AlCl 3 ⋅ AlCl 4− + Cl − ,

pK

[10.4.2]

In the melts, chloride-ion donors were bases and substances increasing AlCl3 concentration were acids. The pK values decrease with the temperature (from 7.1 to 5.0 at temperatures 175 and 400oC, respectively, Na-based melt9) elevation and from Cs to Li (at 400oC pK were 3.8, 5.0, 5.8, 7.4 for Li-, Na-, K- and Cs-based melts, respectively10). The latter effect may be explained from the point of view of “hard” and “soft” acids and bases11-13 - Li-Cl complexes should be more stable than Cs-Cl ones as formed by “hard” base (Cl-) and “more hard” acid (Li+). Dioum, Vedel and Tremillon14 investigated molten KGaX4 (X=Cl, I) as background for acid-base processes. The following acid-base equilibria exist in the pure solvents:

(

)

2GaCl 4− = Ga 2 Cl 7− GaCl 3 ⋅ GaCl 4− + Cl − ,

pK=4.25±0.05

[10.4.3]

GaI 4− = GaI 3 + I − ,

pK=2.6±0.05

[10.4.4]

618

Victor Cherginets

Wagner15 reported an electrochemical study of molten NaBF4 at 420oC. The range of acid-base properties varying according to reaction: BF4− = BF3 + F − ,

pK=1.8

[10.4.5]

has been estimated as ~2pF units. The acid-base ranges in ionic melts containing halide complexes are relatively narrow, it means that varying their halide basicity (i.e., equilibrium concentration of the halide ions) may be performed in 2-4 logarithmic units, although such changes are enough to obtain some unstable compounds with the intermediate oxidation degrees.6,7 10.4.1.2 The Lux-Flood definition Oxygen-containing impurities in molten salts are most usual, their effect on technological processes is mainly negative and consists of bonding acidic reagents with the formation of insoluble (suspensions) or slightly dissociated products. Similar reactions result in retarding the main processes caused by the decrease of the equilibrium concentrations of initial reagents, forming oxide inclusions in metals obtained by the electrolysis of melts, inclusions of insoluble particles in single crystals, considerable corrosion of container for crystal growth, etc. Therefore, quantitative studies of reactions with participation of oxide ions in ionic melts are of importance for scientific and engineering purposes. Reactions with the transfer of oxide ions in ionic media are mainly considered as oxoacidity or acid-base equilibria by Lux-Flood:2-5 A + O-2 = B, K

[10.4.6]

The addition of acids (bases) in a melt as it follows from [10.4.6] leads to changes of the oxide ion activity, for the characterization of melt acidities (basicities) Lux2 proposed the oxygen index, or pO, which was similar to pOH (i.e., basicity index) in aqueous solutions: pO ≡ − log aO 2 −

( ≡ − logm ) O2−

[10.4.7]

where: a O 2 - , m O 2 - activity and molarity of oxide ions in the melt, respectively

The measurements of oxygen indices during various reactions in molten media define their thermodynamic characteristics - dissociation constants and solubility products. Most oxoacidity studies were made in nitrate melts, while there were much less communications dealing with the similar studies in higher-melting alkaline chlorides, molten bromides and iodides. There is a considerable scatter of obtained experimental results, which are often in conflict one with another. There were no serious attempts to generalize these data or consider them from the common point of view. 10.4.2 THE FEATURES OF IONIC MELTS AS MEDIA FOR ACID-BASE INTERACTIONS Ionic melts as media for Lux-Flood acid-base reactions may be divided into two types on the base of constitutional (i.e., being a part of main components of the melt) oxygen ions: oxygen-less and oxygen-containing ones. Let us consider some features of acid-base interactions in the mentioned melts.16


619

10.4.2.1 Oxygen-less media There are no oxide ions in the composition of these melts, therefore the pure melts cannot possess oxide-donor properties. Real oxygen-less melts contain small amounts of O2- owing to inevitable ingress of oxygen-containing impurities into the melt, but oxide ion concentration in the “pure” melts is variable, depending on the concentration and acid-base character of impurities. For example, even considerable amounts of sulfates in melts do not create appreciable O2- concentration, carbonate ion dissociation is substantially stronger, hydroxide ions may be referred to most strongly dissociated Lux bases. In “pure” oxygen-less melts oxygen index pO is usually in the range 3 to 4.5. The employment of strongest purifying agents (HCl, CCl4, etc.) does not allow to decrease this concentration essentially, the latter being in “an unavoidable harm” causing errors at the quantitative investigations, especially when small amounts of acid or bases are studied. But, in some cases, oxygen index for such a “pure” melt has been used as the internal standard for construction of acidity scales.17-20 Quantitative studies of different Lux-Flood acids and bases in ionic melts were performed by two main ways: the construction of empirical acidity scales to estimate relative acidic strength of the substances and the determination of acid-base equilibria constants using potentiometric titration techniques. The first approach has been proposed in the classic work of Lux2 who obtained the acidity scale for the equimolar mixture of potassium and sodium sulfates. Although this work was the basis for series of later studies, the results cannot be considered as undoubted ones. Addition of acids (bases) to the mentioned melt led to increasing (decreasing) e.m.f. of cell with the oxide-selective electrode vs. the corresponding magnitude in the neutral melt. Then, the e.m.f. value shifted progressively to that for the neutral melt because of SO3 (Na2O) evaporation from the acidic (basic) melt. Therefore, Lux extrapolated values of e.m.f. to the point of acid (base) addition to the melt. This resulted in the decrease of the data2 accuracy. The empirical acidity scales17-20 give some information about strength of the acids and bases in melts studied. The principal error related to the term “acidity scale” in the case of oxygen-less melts is because it is not connected with the melt properties. Therefore, the values obtained could not be considered as the quantitative characteristics of the oxygen-less melts. The acidity scale length in a solvent is believed to be the interval (measured in acidity index units) between standard solutions of strong acid and base. If the solvent possesses its own acid-base autodissociation equilibrium then a substance creating unit concentration of acid (base) of solvent in the standard solution should be considered as the strongest acid (base). Addition of stronger acids (bases) should not result in extension of the acidity scale because of the known phenomenon of leveling acidic and basic properties by the solvent.21 Oxygen-less melts do not possess acid-base equilibrium, therefore, values of the basicity index, in the “pure” melts may have any reasonable value. 10.4.2.2 Oxygen-containing melts Acid-base processes in oxygen-containing melts are more complex than those in oxygen-less ones, since they are accompanied by competitive equilibria of own acid-base autodissociation of the melt-solvent. The coexistence of acidity and basicity “carriers” into melt is the characteristic feature of oxygen-containing melts making them similar to low-temperature molecular solvents with own acid-base equilibrium. But, there exist some principal features due to relatively high temperatures of the liquid state. Own acids of the melts are often unstable or volatile,2,22 therefore, acidic solutions,

620

Victor Cherginets

as a rule, lose acids because of their evaporation2 or decomposition,22 the oxygen index of such melts is shifted progressively to that of the neutral melt. Hence, in oxygen-containing melts it is possible to observe not only the leveling of acidic properties but also an upper limit of acidity. CrO3 and MoO3 have been found23 to breakdown the nitrate melt KNO3 with the evaporation of NO2. It should be noted, that the temperature elevation leads to the efficient shrinkage of the acidic region in the melt. The interactions between the oxides and the melt have been assumed to result in the formation of nitrogen (V) oxide, N2O5, and corresponding oxoanions. Since the acids studied (CrO3, MoO3) may be referred to as very strong acids, it may be supposed that reaction: 2CrO 3 + NO 3− = Cr2 O 72 − + NO +2

[10.4.8]

should be completely shifted to the right and the strength of the acid is determined by acidic properties of nitronium cation, NO2+. The stability of the latter is low and after achieving a certain concentration its reaction with the melt anions becomes intensive enough: 1 NO +2 + NO 3− = 2NO 2 ↑ + O 2 2

[10.4.9]

The so-called kinetic methods of melt acidity determination, which will be considered below, are based just on this reaction. The reaction [10.4.9] leads to the decrease of nitronium concentration and the rate of NO2 emission decreases until essentially constant acid concentration is determined by a sequence of consecutive measurements. This concentration is the “upper limit” of acidity of the melt. The thermal dependence of upper limit of acidity23 in nitrate melts can be easily explained on the basis of the increase of the melt temperature, which leads not only to the reduction of nitronium stability but also to the elevation of process [10.4.9] rate. Sulphate melts have the upper limit of acidity too, it seems connected with limited and low solubility of SO3 at elevated temperatures, such assumption may be confirmed by results.2 10.4.2.3 The effect of the ionic solvent composition on acid-base equilibria The equilibrium parameters of Lux acid-base reactions in ionic media (solubility products of oxides and acid-base equilibrium constants) are essentially affected by the acidic properties of the molten alkaline halide mixtures, i.e., they are dependent on the constituent cation acidities. Therefore, one should consider the reverse problem - the estimation of the basicity indices of ionic melts on the basis of the calculated equilibrium constants. The default acid-base processes are actually a superimposed effect of interactions between oxide-ions formed and the most acidic cations of the melt: iMe m+ + O 2 − ↔ Me i O i⋅ m − 2 ,

Kl,i

[10.4.10]

where: Mem+

the most acidic cation of the solvent

The increase of melt acidity (Kl,i) leads the shift of interactions [10.4.10] to the right. The distribution of oxide ions added to the melt between different complexes with the melt cations may be presented by the following equation (N is mole fraction, m - molarity):


621

n   i NO0 2 − = NO 2 −  1 + ∑ K l, i NMe +    i =1

[10.4.11]

It may be seen that the ratio “free oxide-ion/total oxide-ion” is the constant which may be designated as Il:24 I l = NO 2 − / NO0 2 −

(= m

O

2−

/ mO0 2 −

) = 1/  1+ ∑ K n

i =1

l, i

 i NMe +  

[10.4.12]

Now let us to estimate the equilibrium molarity of the constituent acidic cations in the melt, e.g., the eutectic KCl-LiCl melt (0.4:0.6) contains ~8.5 mole of Li+ per 1 kg. Usually the ionic complexes in melts are characterized by the coordination number ~ 4-6.21 For the solution of O2- of the 0.1 mole/kg concentration, the maximum possible quantity of fixed Li+ concentration may be estimated as 0.4-0.6 mole/kg, i.e., efficiently lower than 8.5. In this case the change of actual Li+ concentration is approximately equal to 5-7% and mMe+ in this case may be suggested as constant. Therefore, for each melt the sum in the denominator of [10.4.12] is the constant reflecting its acidic properties. So, pIl= -logIl is a measure of melt acidities and may be denoted as the “oxobasicity index” of the melt. Since the determination of the “absolute” concentration of free O2- is practically impossible one should choose the “standard melt”, for which Il is conditionally equal to 1 and pIl=0. It is reasonable to choose the equimolar mixture KCl-NaCl as the “standard melt”, since this melt is most frequently investigated. Further, one should choose “standard equilibria” and formulate the non-thermodynamic assumptions which usually postulate that the constant of the “standard equilibrium” calculated using “absolute” oxide ion concentrations remains the same for all other melts. Now let us consider possible variants of such “standard equilibria” and conditions of their use. One of the first attempts to define and estimate the oxoacidity parameters of ionic (chloride-based) melts was connected with the studies of the equilibrium H2 O gas + 2Cl − ↔ 2HCl gas + O 2 −

[10.4.13]

in different chloride mixtures.25,26 Here the partial pressures of H2O and HCl are known and the oxide ion concentration is calculated on the base of the calibration data as m 0O 2 - ( N0O 2 - ). The equilibrium constant of [10.4.13] may be represented in terms of the “absolute” mole fractions of O2- and Il by: K1 =

(

2 p HCl NO 2 − I l−1

p H 2O N

2 Cl −

) =K

−1 KCl − NaCl l

I

[10.4.14]

and, consequently: pK l = pK KCl − NaCl − pI l Combes et al.25 introduced the oxo-acidity function, Ω:

[10.4.15]

622

Victor Cherginets

Ω = 14 − pK l + pO 2 −

[10.4.16]

taking into account [10.4.15] eq. [10.4.16] Ω = (14 − pK KCl − NaCl ) + pI l + pO 2 −

[10.4.17]

where (14 - pKKCl-NaCl) is the Ω value in KCl-NaCl standard solution (pO=0) and pO2- is the instrumental pO in the solvent studied. Therefore, the shift of pO scales vs. KCl-NaCl depends only on the Il value. On the basis of [10.4.13] it has been estimated as +7-8 log units (Figure 325) for KCl-LiCl at 1000K. Homogeneous Lux acid-base equilibria of type of [10.4.6] have not been earlier considered as available for estimations of oxoacidity indices. However, pK for these reactions may be written as pK1 = − log

(

NB

NA NO 2 − I l−1

)

= pK KCl − NaCl + pI l

[10.4.18]

For the use of this type of equilibrium it is necessary to make the following non-thermodynamic assumption:  γB   γA

  γB   =    l  γ A  KCl − NaCl

[10.4.19]

where: γ

corresponding activity coefficients of the acid and the base in solvent “l” and KCl-NaCl

The use of [10.4.16] where A and B are anion acid and base for estimations of the oxobasicity indices may be justified because the acid and the conjugated base are negatively charged (in the pair Cr2O7-2/CrO4-2 they are of the same charge). Finally, let us consider the usability of the oxide solubility data MeOs = Me 2+ + O 2 −

[10.4.20]

for pIl estimations. The solubility product value, P, may be presented as

(

)

PMeO, l = NMe 2 + NO 2 − I l−1 = PMeO, KCl − NaCl I l−1

[10.4.21]

pPMeO, l = pPMeO, KCl − NaCl − pI l

[10.4.22]

or

Metal cations in molten halides form halide complexes. The reaction [10.4.20] for chloride melts (NCl- =1) suggests that the distribution of Me2+ between different complexes MeCln2-n remains unchanged (similarly to eq. [10.4.10]-[10.4.12] for the oxide ion distribution). It is clear that the solubility data may be used for estimations of the oxoacidity indices only in melts with the same anion composition. The anion changes cause errors since Me2+ in eq. [10.4.21] is referred to essentially different acids, e.g., halide complexes MeCl42- and


623

MeBr42- in the case of chloride and bromide melts. Therefore, regardless of the cation composition, the oxide solubilities are essentially affected by anion composition of a melt.27 Equations [10.4.14], [10.4.18] and [10.4.22] estimate pIKCl-LiCl vs. KCl-NaCl as 7-8,25 24 3.7 and 3.424, respectively. This estimation is considerably larger than the estimation based on eq. [10.4.18] and [10.4.22]. For Ca2+ based chloride melts, the similar estimation made on the basis of [10.4.14] gave pI~1025 vs. ~4 determined on the basis of [10.4.22].28 So, the oxobasicity indices calculated according to HCl/H2O equilibrium constant are essentially different from those obtained using reactions without reactive gas atmosphere. Probably, this discrepancy may be explained by the features of the water behavior in aprotic ionic melts. It is known that, similar to basic properties, water possesses oxoacidic properties according to the reaction:

(

H2 O + O 2 − ↔ 2OH − or (OH) 2

−2

)

[10.4.23]

Such polynuclear complexes are stable according to spectral data29 because of the H-bonds formation. Three acids (HCl + Me+ + H2O) exist in ionic melts saturated with water and HCl and consequently reaction [10.4.23] depends on the partial pressure of water. In Liand Ca-based chloride melts, retention of water30 and the solubility of H2O is appreciably higher than in e.g. KCl-NaCl. Apparently, even the simplest reaction [10.4.13] is actually complicated by additional interactions favoring the fixation of “free” oxide ions. Values of oxobasicity indices calculated from eq. [10.4.13] are thus somewhat overestimated. The oxobasicity index for KCl-LiCl (0.4:0.6) compared with KCl-NaCl at 700oC lies within the range 3.4-3.7 log units. 10.4.3 METHODS FOR ESTIMATIONS OF ACIDITIES OF SOLUTIONS BASED ON IONIC MELTS The literature data show that many methods were used for oxoacidity studies and estimation of the acidic properties of melts. One simplest method31 involves indicator. Acid-base indicators usually employed in aqueous solutions for protic acidity measurements have been used for acidity studies in molten KNO3-LiNO3 at 210oC and KSCN at 200oC. The color of indicator solution, relative to acidity, changes during titration of bases (sodium hydroxide or peroxide) by potassium pyrosulfate, K2S2O7: 2HInd + O 2 − = 2Ind − + H 2 O

[10.4.24]

where: HInd Ind-

the protonized form of the acid-base indicator the anionic form of the said indicator

In molten nitrates such transition was observed only for phenolphthalein (yellow-purple), other indicators seemingly were oxidized by the melt (the conclusion made31 was “insoluble in the melt”). In molten KSCN not possessing oxidizing properties color transitions were observed for all indicators used (methyl red, thymolphthalein, etc.). The employment of indicators for acidity estimation is limited mainly by their thermal instability and tendency to oxidize at high temperatures. Since ionic melts, as a rule, have no constitutional water, the reverse transition of indicator into the protonized form [10.4.24] is hardly possible (the solution of phenolphthalein became yellow at the reverse transition “base-acid”31).

624

Victor Cherginets

Therefore, the use of indicator method is possible only at low-temperatures (250oC) and in non-oxidizing ionic melts. Spectral methods were also used for the melt basicity estimation.32,33 The scale of “optical” basicity for metallurgical slugs with respect to CaO (standard) has been constructed.32 The “optical” basicity, Λ, was estimated from spectral line shifts (the transition 1S0 → 3P1) for Tl+, Pb2+, Bi3+ cations in the basic medium. Ionic melts Na2O-B2O3-Al2O3, Na2O-B2O3SiO2, Na2O-SiO2-Al2O3 have been investigated by X-ray fluorescent spectroscopy,32 relative oxide acidities decrease in the sequence B2O3>SiO2>Al2O3. The use of spectral methods is based on studies of cooled (quenched) samples, such a routine may distort results because of inconsistency between the solution temperature and that of acidity estimations. Data on the melt basicity estimations by the determination of acidic gases (SO3, CO2, H2O) solubilities are presented elsewhere.34,35 Sulphur trioxide solubility in molten sodium phosphate was determined by thermogravimetric analysis,34 the correlation between the melt basicity and the SO3 solubility was found. The use of CO2 and H2O for basicity estimation was described elsewhere.35 The similar methods may be used only for basic melts since acids displace the acidic gas from the melt. Furthermore, the interaction products of the acidic gas with the melt may be relatively stable, especially in basic solutions, and their formation leads to irreversible changes of the melt properties. A “kinetic method” was used for Lux acidity studies in molten nitrates. The interaction between the nitrate melts KNO3-NaNO3 and potassium pyrosulfate was studied.36,37 S 2 O 72 − + NO 3− = 2SO 42 − + NO 2+

[10.4.25]

The formed nitronium cation, NO2+, reacted with NO3- according to [10.4.9]. The latter stage is considerably slower than [10.4.25], hence, its rate allowed to estimate the melt acidity, which can be presented by the following sum: TA = [S2O72-] + [NO2+].36 The acid-base interactions between Cr2O72- and chlorate ions, ClO3-, in molten KNO3-NaNO3 are described by the following equations.38,39 Cr2 O 72 − + ClO 3− = 2CrO 42 − + ClO +2

[10.4.26]

ClO +2 + Cl − → Cl 2 ↑ +O 2 ↑

[10.4.27]

From two above processes, reaction [10.4.27] is the limiting stage, the rate of the chlorine evolution is proportional to the total acidity of the melt: TA = [Cr2O72-] + [ClO2+]. The reaction of bromate ions, BrO3-, with potassium dichromate in the nitrate melt has been studied.38 Several studies36-40 gave estimate of relative acidities of the oxo-cations. The acidities increased in the sequence BrO2+TiO2>ZrO2>Al2O3>MgO. Potentiometric method65-67 used metal-oxide electrodes (Nb|Nb2O5, Ta|Ta2O5, Zr|ZrO2) for oxoacidity studies in the nitrate melt. The empirical acidity scale in molten potassium nitrate at 350oC was constructed: NH4VO3>NaPO3>NaH2PO4>K2Cr2O7>K2HPO4> Na4P2O7>NaHAsO4>K2CO3>Na2O2. This sequence includes some acids seldom existing in molten salts. NH4VO3 at 350oC should be completely transformed into V2O5 with a partial reduction of the latter by ammonia. Acidic salts mentioned were decomposed to corresponding pyro- and meta-salts. Therefore the scale obtained contains a number of errors. 10.4.4.2 Sulphate melts Sulphate melts are referred to melts with the own acid-base dissociation equilibrium: SO 42 − = SO 3, l ↑ +O l2 −

[10.4.37]

628

Victor Cherginets

There are rather few contributions devoted to the Lux acidity studies in the similar melts. This may be explained by relatively high melting points both of single sulfates and their eutectic mixtures. The weak stability of acidic solutions in sulphate melts can also be considered as a reason why these melts are studied insufficiently. Indeed, reaction [10.4.37] should result in SO3 formation in the melt. Reference data,68,69 however, show that decomposition temperatures of pyrosulfates (i.e., complexes SO3(SO42-) do not exceed 460oC, in dissolved state stability of solvated SO3 should be even lower. The absence of SO3 partial pressure over the melt gives rise to removing SO3 from the latter and, since the equilibrium state of [10.4.37] is not attained, SO3 should be removed completely. Among sulphate melts the ternary eutectic mixture K2SO4-Li2SO4-Na2SO4 (0.135:0.78:0.085) with the lowest melting point 512oC70 may be most promising for the acidity studies since it is possible to perform studies at the temperatures significantly lower than in the classic work of Lux.2 Under these conditions stability of acidic and basic solutions increases. Lux2 reported the oxygen electrode reversibility investigation and construction of empirical acidity scale in the molten eutectic mixture K2SO4-Na2SO4 at 800oC. The thermal stability of pyrosulfates has been shown by Flood3 to decrease with melt cation acidity increasing. Kaneko and Kojima71,72 investigated acid-base properties of different solutions in the molten K2SO4-Li2SO4-Na2SO4 at 550oC; the empirical acidity scale length was approximately 10 pO units between 0.01 mole/kg solutions of S2O72- and O22- ions. From the latter data, the acid-base interval of this solvent can be easily estimated as the length of acidity scale between standard solution - ~14 pO units. The pO range of VO2+ stability has been determined to be 8.5 to 10.6. To complete the consideration of investigations in sulphate melts the work73 should be mentioned, it is concerned with potentiometric studies of the acidic properties of MoO3, Cr2O72-, PO3-, P2O74-, V2O5 in molten K2SO4-Li2SO4-Na2SO4 at 625oC. Acids mentioned were neutralized by sodium carbonate. The potential (pO) drop in equivalence points was observed at “acid-base” ratio 1:3 for vanadium oxide and 1:1 for all other Lux acids. Values of equilibrium constants were not presented. 10.4.4.3 Silicate melts Studies performed in molten silicates are of importance for applied purposes, since systems under consideration are widely used in various industries (slugs, glasses, etc.). Flood et al.74-76 studied the acidity dependence of silicate melts upon their composition. A system PbO-SiO2 have been studied in the range of SiO2 concentrations of NSiO 2 from 0 to 0.6 at 1100-1200oC. The coexistence of (SiO3)36- and (SiO2.5)66- polyanions together with the ordinary orthosilicate ions, SiO44- has been shown. The basicities of molten glasses vs. Na2O/SiO2 ratios were studied75 by the potentiometric method with the use of the gas oxygen electrode Ag(O2) in concentration cells. The decrease of the mentioned ratio has been stated to cause a reduction of the basicity, while magnesium oxide has been found to have no effect on the melt basicity. The gas oxygen electrode reversibility in molten PbO-SiO2, Na2O-CaO-SiO2, MeO-PbO-SiO2, where Me2+ - alkaline earth cation, has been investigated.77 The electrode Pt(O2) has been shown to be reversible to O2-. The use of a membrane oxygen electrode Pt(O2)|ZrO2 for molten silicate acidity control during glass making has been described.78 Studies of basicities in molten Na2O-Al2O3-SiO2 using a potentiometric method79 showed that the acidic properties of Al2O3 were weaker, then SiO2 ones. Processes of glass corrosion in molten alkaline earth nitrates have been studied for glass compositions Na2O⋅xAl2O3⋅2SiO2, where x was in the range from 0 to 0.4.80 The degree of interaction between the melt and immersed glass in-


629

creased with the reduction of the alkaline earth cation radius. The effect of the cation charge and the radius on the basicity of lead meta-silicate melt at 800, 850, 900oC by additions of Tl2O, PbO, CdO, ZnO and Bi2O3 has been studied by potentiometry.81 10.4.4.4 The equimolar mixture KCl-NaCl This melt is the most investigated among chloride based melts. Some oxoacidity studies were performed by Shapoval et al.82-87 Main purposes of these works were to investigate oxygen electrode reversibility and to obtain equilibria constants for acid-base reactions including oxo-compounds of CrVI, MoVI and WVI at 700oC. Equilibrium constants of acid-base interactions for CrO3 and PO3- have been determined.82-84 The titration of the former substance proceeds in two stages: −1

. 88 ×10 . 4 ×10 → Cr2 O 72 − 1 → CrO 42 − CrO 3  2 3

[10.4.38]

Sodium metaphosphate is two-basic acid too: 2. 88±1. 2 ) ×10 2. 5±1. 4 ) ×10 → P2 O 74 −  ( → PO 43− PO 3−  ( 3

1

[10.4.39]

The excess of titrant, O2-, has been found to cause the formation of basic phosphates with assumed composition PO3-⋅2O2-. When Na2CO3 was used as titrant, there was no formation of the mentioned basic products. Ba2+ and Li+ cations have been shown to possess appreciable acidic properties, corresponding constant values were estimated as 8.1×101 and 3.53×102. An investigation of acidic properties of MoO3 was performed:85 ×10 ±880 MoO 3  5 → Mo 2 O 72 −  880  → MoO 42 − 4

[10.4.40]

Acidic properties of oxo-compounds of WVI have been studied.86,87 The titration of WO3 in molten chlorides was a two-stage process: ±390 ±880  → W 2 O 72 −  880  → WO 42 − WO 3 1460

[10.4.41]

Results formed the base for the following electrochemical studies of electroreduction processes of Group VI metals.88-90 These studies developed theoretical bases and principles to control electrochemical processes of metals and their compounds (carbides, borides, silicides) deposition from ionic melts.81 The acid-base equilibria in the scheelite (CaWO4) solutions in the said chloride mixture at 1000K were investigated.92 CaO solubility in molten KCl-NaCl has been determined to be about 0.084 mol%, that of scheelite was 10-3.5 mole/kg. Equilibrium constants for acid-base reactions with WO3 participation have been determined, too:  → WO 42 − WO 3 10 10

[10.4.42]

The reverse titration has been assumed by authors92 to result in the formation of polyanion W3O102-, pK=12.7. The titration curve was irreversible. This result can be easily explained taking into account that WO3 is unstable in strong acidic chloride melts and removed from as WO2Cl2. Therefore, additions of WO3 into the acidic melt should lead to un-

630

Victor Cherginets

controlled titrant loss affecting experimental results. Ditto referred to all results obtained by the reverse titration by MoO3 and CrO3 as the equivalence point. Acidic properties of PV oxocompounds in the chloride melt at 700oC were investigated using a membrane oxygen electrode Ni,NiO|ZrO2.93 Polyphosphates of compositions with Na2O-P2O5 ratio from 1.67 to 3 formed in the acidic region: Pn O 3−n2+(1n+2 ) + PO 43 − = Pn+1O 3−n( +n5+5 )

[10.4.43]

The corresponding constants were estimated. Polyphosphate solubilities were determined by a cryoscopic method. The titration of V2O5 proceed in two stages:94 = −9. 3±0. 3 = −8. 3±0. 3 V2 O 5  pK  → VO 3−  pK  → V2 O 74 −

[10.4.44]

Some results of oxoacidity studies in molten KCl-NaCl were obtained in our works,95-99 they are presented in Table 10.4.1. Table 10.4.1. The Lux acid-base equilibrium constants in molten KCl-NaCl and NaI (at the confidence level 0.95) [After references 95,96,98] -pK

Equilibrium

m

N

mol%

8.01±0.1

10.67

6.67

5.93±0.1

7.26

5.26

KCl-NaCl 2PO + O 3

− 3

PO + O

2−

2−

− 3

Cr2 O

2− 7

2CrO

= PO

= [ PO ⋅ O ]

7.24±0.1

-

-

2−

2− 4

3− 4

8.41±0.1

9.77

7.77

7.18±0.1

7.18

7.18

1.60±0.2

4.26

0.26

8.32±0.2

9.65

7.65

9.71±0.3

11.37

7.37

9.31±0.2

10.64

8.64

10.67±0.5

13.33

9.33

4.82±0.1

2.16

4.16

2.37±0.2

3.17

1.70

6.95±0.2

6.95

6.95

12.23±0.1

14.89

10.89

12.30±0.1

14.96

10.96

13.88±0.5

19.20

11.20

4.18±0.4

6.84

2.84

= CrO

+O

2− 4

2−

+O

MoO3 + 2O 2−

2− 7

2−

B4 O − 2

+O

BO + O

2−

V2 O5 + O

2−

V2 O5 + 3O

2−

V2 O5 + 5O

2−

2GeO2 + O

2−

2− 4

2−

2− 4

2− 4 2−

= [WO ⋅ O ] = 4BO

− 2

3− 3

= 2VO

V2 O5 + 2O

2−

2− 4

= BO

2−

2− 4

= [MoO ⋅ O ]

= WO

WO3 + 2O

2− 4

= [2CrO ⋅ O ]

= MoO

2−

2−

2−

= 2CrO

2−

2−

MoO3 + O WO3 + O

3− 4

2−

PO + 2O CrO3 + O

= P2 O

4− 7

− 3

= V2 O

4− 7

= 2VO

3− 4 3− 4

2−

= 2[VO ⋅ O ] = Ge2 O

2− 5


Equilibrium

631

-pK m

N

mol%

NaI − 3

2VO + O

2−

4− 7

2−

V2 O

B4 O

2− 7

+O

+O

2−

= V2 O

4− 7

= 2VO

5.40±0.3

7.10

4.10

3− 4

1.68±0.3

1.68

1.68

− 2

5.02±0.8

3.37

5.37

= 4BO

The results included in the Table 10.4.1 are in good agreement with the similar studies, performed using a membrane oxygen electrode Ni,NiO|ZrO2. The use of the gas platinum-oxygen electrode resulted in significant underrating of constant values (of order of 3-4 pK units). 10.4.4.5 Other alkaline halide melts Some studies of acid-base equilibria in molten alkaline halides have been performed by Rybkin et al.17-20 The empirical acidity scales have been constructed on the base of potentiometric studies in molten KCl17,18 and CsI.19 The practical significance of the mentioned works consists in the proposition to use some buffer solutions - SO42-/S2O82-, WO42-/W2O72-, PO43-/P2O74- as reference standards for the indicator electrode calibration and determinations of pO in molten alkaline halides. Rybkin and Banik20 obtained the empirical acidity scale in molten NaI at 700oC. The basicities of CO32- and OH- were close which is incorrect since 0.01 mole/kg solutions were used for the scale constructing while equations 1 1 OH − = H2 O + O 2 − 2 2

[10.4.45]

CO 32 − = CO 2 ↑ + O 2 −

[10.4.46]

showed that these additions were not equivalent. CO32- concentration recalculated to oxide ions was twice as high as OH concentration. Indeed, in the later work these authors100,101 showed that hydroxide ion was stronger base than carbonate ion. We have performed a study of acidic properties of boron (III) and vanadium (V) oxo-compounds in molten NaI at 700oC98 (see Table 10.4.1). A comparative study of strength of Lux bases OH-, CO32-, SO42- was determined100,101 by potentiometric titration using sodium pyrophosphate as acid. Two moles of the first base may be neutralized by 1 mole of P4O72-, while two other moles react with the acid in ratio 1:1. On the base of e.m.f. drop magnitude at the equivalence point, the bases have been arranged in sequence OH->CO32->SO42- of basicity decrease. The equilibrium constants were not estimated.100,101 There is no correlation17,19 between substances entered into melt and those really existing in it. E.g., potassium nitrate and nitrite17,18 are decomposed to K2O at temperatures considerably lower than the temperature of the experiment (700oC). Ditto17,18 referred to K2S2O7 and Na2S2O8 as acidic phosphates.19 In the latter work it has been found that pyrophosphate acidity is larger than that for metaphosphate. However, titration of phosphorus (V) oxo-compounds in melts runs according to the scheme [10.4.39]. The acidity decreases,

632

Victor Cherginets

and, hence, pyrophosphate acidity should be lower, since in pair “metaphosphate-pyrophosphate” the latter is the conjugated base. The absence of correlation between pH in aqueous solutions and pO in melts for the same substance is often noted , this conclusion is made mainly from data for oxocompounds of PV. This fact is due to the following reason. The stages of phosphoric acid neutralisation in aqueous solutions are: −

−

−

 → H2 PO 4− OH  → HPO 42 − OH  → PO 43 − H3 PO 4 OH

[10.4.47]

In melts, a similar sequence can be presented as 1 1 / 2O 2 − / 2O 2 − / 2O 2 − P2 O 5 1 → PO 3− 1 → P2 O 74 − 1 → PO 43− 2 2

[10.4.48]

Hence, correlations should be found between the anion in aqueous solution and corresponding anhydro-acid in melt. NaPO3 and Na4P2O7 may be easily obtained by the calcination of the corresponding acidic salts, but reverse processes do not take place in aqueous media and pyro- and metaphosphate exist in water as salts of stable acids. For arsenates which are more prone to hydrolysis similar correlation should take place. The absence of the correlation is caused mainly by kinetic limitations. 10.4.5 REACTIONS OF MELTS WITH GASEOUS ACIDS AND BASES The gases present in the atmosphere over molten salts can react both with their principal components and with impurities. In reactions of the first kind, which include, in particular, hydrolysis, the gas (water) behaves as a Lux base since its action results in the increase of O2- concentration. In reactions of the second type, which are usually employed for the purification of melts, the gas has acidic properties. 10.4.5.1 High-temperature hydrolysis of molten halides These reactions were investigated mainly for alkaline halides and their mixtures. Hanf and Sole102 studied the hydrolysis of solid and molten NaCl in the temperature range 600-950oC by so-called “dynamic method” which consisted of passing inert gas (N2) containing water vapor through a layer of solid or fused NaCl. The developed routine determined the equilibrium constant of the following reaction: NaCl s,1 + H2 O gas = NaOHs,1 + HCl gas

[10.4.49]

The values of the equilibrium constants (log K) of [10.4.49] for solid and liquid sodium chloride were in the ranges -8 to -6 (650-800oC) and -6 to -5 (800-900oC). The hydrolysis of NaCl occurs to a slight extent, although the rate of HCl evolution exceeded the expected value because of dissolution of NaOH formed in NaCl. An evident disadvantage of the above method102 is assumption that the partial pressure of HCl in the gas phase is equal to the equilibrium pressure. The same method was used to study the high-temperature hydrolysis and oxidation of sodium iodide.103 According to the estimates, the heats of the reactions: NaI + H2 O = NaOH + HI

[10.4.50]


NaI + 1 / 2 O 2 = 1 / 2 Na 2 O 2 + 1 / 2 I 2

633

[10.4.51]

are 32.8±2 and 13.9±1 kJ/mole for solid NaI and 30.6±2 and 9.3±2 kJ/mole for liquid one. These values suggest that NaI has a greater tendency to oxidation than to hydrolysis. This conclusions cannot be adopted without stipulation, because it is based on values ∆H and not ∆G, although only the latter values determine the direction of a reversible reactions at constant p and T. A simple but unquestionable method has been used to investigate high-temperature hydrolysis of molten KCl-NaCl.104 A mixture of HCl and H2O obtained by passing an inert gas through aqueous solutions at a definite concentration was passed into the melt. Measurements of the equilibrium O2- concentration by potentiometric method allowed to calculate the equilibrium constants of [10.4.13] as pK=55.3×103T-1-40.2. At 1000, pK=15.1, it means that equilibrium [10.4.13] in molten KCl-NaCl is displaced to the left. A similar method has been used to study the hydrolysis of the eutectic KCl-LiCl at 500oC (pK=9.77±0.4).105 The hydrolysis of the chloride melt is thermodynamically unfavorable and is completely suppressed in the presence of bases (even at a concentration of ~10-3 mole/kg). The potentiometric cells:

(H2 O + H2 )Au|N MeOH + (1 − N)MeX ||MeX |C(X 2 )

[10.4.52]

were used to investigate the hydrolysis of individual alkaline halide melts (with the exception of the lithium salts).106 The e.m.f. of cell [10.4.52] is related to the concentration of the reactants by the expression: E =E 0 +

1/ 2 1/ 2 RT PH 2 PX 2 ln a MeOH F PH 2O

[10.4.53]

where PH1/22 and PX1/22 are partial pressures of hydrogen and halogen over the melt, respectively. The logarithms of the equilibrium constants 0 PH1/22 PX1/22 ∆G HX F ln K = + ln E− a MeOH RT RT PH 2O

[10.4.54]

are negative for all the melts studied, which indicates that the hydrolysis of alkali metal halide melts is thermodynamically unfavorable. The trend towards hydrolysis of melts of individual alkaline halides diminishes with the increase in the radii of both cation and anion. According to the results of Smirnov et al.106 the solutions of alkali metal hydroxides in the corresponding individual halides are close to ideality. 10.4.5.2 The processes of removal of oxide admixtures from melts The studies of the elimination of oxygen containing impurities from ionic melts are of an applied character and are performed in melts having industrial applications. Various

634

Victor Cherginets

halogenating agents, usually hydrogen halides or the halogens themselves, are used to eliminate the impurities:107-111 2HX ↑ +O 2 − = H2 O ↑ +2X −

[10.4.55]

X 2 ↑ + O 2 − = 1 / 2O 2 ↑ + 2X −

[10.4.56]

However, the water formed in reaction [10.4.55] exhibits the properties of Lux acid owing to process opposite to [10.4.45] and this retards the purification process. The cessation of the HX effect may lead to partial hydrolysis, because reactions of type of [10.4.55] are reversible. The shift of [10.4.55] to the left is also promoted by the thermal instability of HBr and HI. The effectiveness of the purification of melts based on lithium salts by hydrogen halides is much lower. Water dissolves in the KCl-LiCl melt in appreciable amounts and is firmly retained at temperatures up to 400oC.30 When dry HCl is passed for 1 h, removal of H2O is incomplete. The effectiveness of the purification by the halogens depends on their oxidation-reduction potentials and decreases from chlorine to iodine. In addition, the latter can disproportionate: 3I 2 ↑ +O 2 − = 5I − + IO 3−

[10.4.57]

However, the addition of NH4X is recommended only for the drying and heat treatment of the initial components and its use for industrial purification of fused salts is undesirable because it is much easier to ensure the continuous supply of the corresponding hydrogen halide in the melt being purified. An additional lack of this purifying agent results in evolution of gaseous ammonia as one of the finishing products. It is believed that the most convenient method of the elimination of oxygen-containing impurities from alkali metal halide melts is carbohalogenation.112-118 The reduction of O2- concentration under action of CCl4, Cl2, C (C is acetylene carbon black)+Cl2 and COCl2 may be explained by the occurrence of the following reactions: CCl 4 ↑ +2O 2 − = 4Cl − + CO 2 ↑

[10.4.58]

C + Cl 2 ↑ +O 2 − = 2Cl − + CO ↑

[10.4.59]

COCl 2 ↑ +O 2 − 2Cl − + CO 2 ↑

[10.4.60]

Thermodynamic analysis of processes [10.4.58]-[10.4.60] showed that the effectiveness of the purification is in all cases approximately the same,113 but reaction [10.4.58] has the advantage that highly toxic reagents (Cl2, COCl2) are not used in it. For this reason, CCl4 has been used in virtually all subsequent studies involving the carbochlorination of chloride melts.114-118 It has been suggested that CHBr3, CBr4, C2H5Br and CHI3 may be used for the removal of oxygen-containing impurities from bromide and iodide melts.119 A significant advantage of carbohalogenation is that it leads to an appreciable decrease in the concentration of transition metal cations in the melt being purified. However, in the purification of melts by halo-


635

gen-substituted compounds with a relatively low halogen content, carbon (carbon black) accumulates in the melt and is displaced by the crystallization front when single crystals are grown.112 The presence of a suspension of carbon in alkali metal halide melts is not always desirable and this method cannot therefore be recognized as universal. The carbohalogenation of melts results in formation of CO2 as one of products of [10.4.58]-[10.4.60] processes, carbon dioxide reacts with remaining oxide ions and induces the accumulation of fairly stable carbonate-ions. The latter is the main form in which oxygen-containing impurities exist in melts purified in this way. The purification threshold of carbohalogenation is estimated as 1-2×10-4 mole/kg of O2-.24 The use of halo-derivatives of silicon for the purification of iodide melts120 is based on the reaction: SiX 4 ↑ +2O 2 − = 4X − + SiO 2 ↓

[10.4.61]

the product of [10.4.61] is silicon dioxide, which is separated from the melt as a consequence of the difference between the densities. Thermodynamic analysis showed121 that the effectiveness of the purification of iodide melts diminishes in the sequence SiI4>HI>I2. The use of silicon halides for the purification of the melts used in the growing single crystals does not lead to the appearance of additional impurities, because the processes are carried out in quartz (i.e., SiO2) containers. Although the methods of purification described are fairly effective in most cases, some kind of oxygen-containing and cationic impurities, characteristic of each method, always remains in the melt. The method of purification must therefore be selected depending on the aims of the subsequent application of the melt. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

G.N. Lewis, Valence and the structure of atoms and molecules, S.n., New York, 1923. H. Lux, Z. Elektrochem., 45, 303(1939). H. Flood and T. Forland, Acta Chem.Scand., 1, 592 (1947). H. Flood and T. Forland, Acta Chem.Scand., 1, 781 (1947). H. Flood, T. Forland and B. Roald, Acta Chem.Scand., 1, 790 (1947). N.J. Bjerrum, C.R. Roston and G.P. Smith, Inorg.Chem., 6, 1162 (1967). N.J. Bjerrum and G.P. Smith, Inorg.Chem., 7, 2528 (1968). B. Tremillon and G. Letisse, J. Electroanal. Chem., 17, 371 (1968). G. Torsi and G. Mamantov, J. Electrocanal. Chem., 30, 193 (1971). G. Torsi and G. Mamantov, Inorg. Chem., 11, 1439 (1972). R. G. Pearson, J. Amer. Chem. Soc., 85, 3533 (1963). R. G. Pearson, J. Chem. Educ., 45, 581 (1968). R. G. Pearson, J. Chem. Educ., 45, 643 (1968). I. G. Dioum, J. Vedel, and B. Tremillon, J. Electroanal. Chem., 137, 219 (1982). J. F. Wagner, Electrochemical study in molten sodium fluoroborate at 420oC, Rep. CEA-N-2350, 15(2), Abstr. No15:005221 (1983). V. L. Cherginets, Rus. Chem. Rev., 66, 597 (1997). Yu. F. Rybkin and A. S. Seredenko, Ukr. Khim. Zhurn., 36, 133 (1970). Yu. F. Rybkin and A. S. Seredenko, Ukr. Khim. Zhurn., 40, 137 (1974). Yu. F. Rybkin and V. V. Banik, In Single Crystals and Engineering , No. 2(9) issue, Inst. Single Crystals, Kharkov, 152 (1973). N. N. Ovsyannikova and Yu. F. Rybkin, Ukr. Khim. Zhurn., 42, 151 (1976). J. E. Huheey, Inorganic chemistry. Principles of structure and reactivity, Harper and Row Publishers, New York, 1983. M. Dratovsky and D. Havlichek, Electrochim. Acta, 28, 1761 (1983).

636

23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76

Victor Cherginets A. M. Shams El Din and A. A. El Hosary, J. Electroanal. Chem., 9, 349 (1965). V. L. Cherginets and T. P. Rebrova, Electrochim. Acta, 45, 469 (1999). R. Combes, M. N. Levelut and B. Tremillon, Electrochim. Acta, 23, 1291 (1978). R. Combes, B. Tremillon, F. De Andrade, M. Lopes and H. Ferreira, Anal. Lett., 15A, 1585 (1982). V. L. Cherginets, Electrochim. Acta, 42, 3619 (1997). Y. Castrillejo, A. M. Martinez, G. M. Haarberg, B. Borresen, K. S. Osen and R. Tunold, Electrochim. Acta, 42, 1489 (1997). A. I. Novozhilov and E. I. Ptchelina, Zhurn. neorg. Khim., 22, 2057 (1977). W. J. Burkhardt and J. D. Corbett, J. Amer. Chem. Soc., 79, 6361 (1957). B. J. Brough, D. H. Kerridge and M. Mosley, J. Chem. Soc. (A), N11, 1556 (1966). J. A. Duffy, M. D. Ingram and I. D. Sommerville, J. Chem. Soc. Faraday Trans., 74, 1410 (1978). T. Maekawa and T. Yokokawa, Nippon Kagaku, N6, 900 (1982). A. Kato R. Nishibashi, M. Hagano and I. Mochida, J. Amer. Ceram. Soc., 55, 183 (1972). N. Iwamoto, Youen, N21(3), 287 (1978). F. R. Duke and S. Yamamoto, J. Amer. Chem. Soc., 80, 5061 (1958). F. R. Duke and S. Yamamoto, J. Amer. Chem. Soc., 81, 6378 (1959). J. M. Shlegel, J. Phys. Chem., 69, 3638 (1965). J. M. Shlegel, J. Phys. Chem., 71, 1520 (1967). F. R. Duke, J. M. Shlegel, J. Phys. Chem., 67, 2487 (1963). I. Slama, Coll. Czechoslov. Chem. Commun., 28, 985 (1963). I. Slama, Coll. Czechoslov. Chem. Commun., 28, 1069 (1963). A. M. Shams El Din and A. A. El Hosary, J. Electroanal. Chem., 8, 312 (1964). A. M. Shams El Din and A. A. A. Gerges, J. Inorg. Nucl. Chem., 26, 1537 (1963). A. M. Shams El Din and A. A. A. Gerges, Electrochim. Acta, 9, 123 (1964). A. M. Shams El Din and A. A. El Hosary, J. Electroanal. Chem., 7, 464 (1964). A. M. Shams El Din and A. A. El Hosary, J. Electroanal. Chem., 17, 238 (1968). A. M. Shams El Din, A. A. El Hosary and A. A. A. Gerges, J. Electroanal. Chem., 6, 131 (1963). A. M. Shams El Din, Electrochim. Acta, 7, 285 (1962). A. M. Shams El Din and A. A. A. Gerges, J. Electroanal. Chem., 4, 309 (1962). A. M. Shams El Din and, A. A. El Hosary, J. Electroanal. Chem., 16, 551 (1968). A. M. Shams El Din, A. A. El Hosary and H. D. Taki El Din, Electrochim. Acta, 13, 407 (1968). A. A. El Hosary, M. E. Ibrahim and A. M. Shams El Din, Electrochim. Acta, 24, 645 (1979). A. M. Shams El Din and A. A. El Hosary, Electrochim. Acta, 13, 135 (1968). A. M. Shams El Din and A. A. El Hosary, Electrochim. Acta, 12, 1665 (1967). J. D. Burke and D. H. Kerridge, Electrochim. Acta, 19, 251 (1974). A. M. Shams El Din and A. A. A. Gerges, In Electrochemistry, Pergamon Press, 1964 pp. 562-577. A. M. Shams El Din and A. A. A. Gerges, Electrochim. Acta., 9, 613 (1964). N. Coumert and M. Porthault, J. -C. Merlin, Bull. Soc. Chim. France, 33, 910 (1965). N. Coumert and M. Porthault, J. -C. Merlin, Bull. Soc. Chim. France, 35, 332 (1967). M. Hassanein and N. S. Youssep, Indian J. Chem., A21, 72 (1982). C. Liteanu, E. Cordos and L. Margineanu, Rev. Roumaine Chim., 15, 583 (1970). J. M. Schlegel, J. Chem. Educ., 43, 362 (1966). Y. Hoshino, T. Utsunomiya and O. Abe, Bull. Chem. Soc. Jpn., 54, 135 (1981). A. Baraka, A. Abdel-Razik and A. J. Abdel-Rohman, Surface Technol., 25, 31 (1985). A. Baraka, A. J. Abdel-Rohman and E. A. El-Taher, Mater. Chem. Phys., 9, 447 (1983). A. Baraka, A. J. Abdel-Rohman and E. A. El-Taher, Mater. Chem. Phys., 9, 583 (1983). A. I. Efimov, Properties of inorganic compounds. Handbook, Khimiya, Leningrad, 1983. I. T. Goronovskiy, Yu. P. Nazarenko and E. F. Nekryach, Short handbook on chemistry, Naukova Dumka, Kiev, 1987. G. J. Janz, Molten salts handbook, Academic Press, New York, 1967. Y. Kaneko and H. Kojima, Int. Symp. Molten Salts Chem. Technol., No. 1, 441 (1983) cited Chem. Abstr. 102, Abstr. No. 122309c (1988. ) Y. Kaneko and H. Kojima, Youen, 28, 109 (1985). A. Rahmel, J. Electroanal. Chem., 61, 333 (1975). H. Flood and W. Knapp, J. Amer. Ceram. Soc., 46, 61 (1963). T. Forland and M. Tashiro, Glass Ind., 37, 381 (1956). T. Forland, Glastekn. Tidskr., 17, 35 (1962).


77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120

637

V. I. Minenko,C. M. Petrov and N. S. Ivanova, Izv. vysshikh utchebnykh zavedeniy. Chernaya metallurgiya, N7, 10 (1960). US Patent 4,313,799; Chem. Abstr., 96, 115179g (1982). H. Itoh and T. Yokokawa, Trans. Jpn. Inst. Met., 25, 879 (1984). T. T. Bobrova, V. V. Moiseev, T. V. Permyakova and G. E. Sheshukova, Izv. AN SSSR. Neorg. mater, 9, 1416 (1973). R. Didtschenko and E. G. Rochow, J. Amer. Chem. Soc., 76, 3291 (1954). V. I. Shapoval and O. G. Tsiklauri, Conf. "Phys. chemistry and electrochemistry of molten salts and solid electrolytes", Sverdlovsk, June 5-7, 1973, Part II, Sverdlovsk, 1973 pp. 32-33. V. I. Shapoval, A. S. Avaliani and O. G. Tsiklauri, Soobsch. AN Gruz. SSR, 72, 585 (1973). Yu. K. Delimarsky, V. I. Shapoval,O. G. Tsiklauri and V. A. Vasilenko, Ukr. Khim. Zhurn., 40, 8 (1974). V. I. Shapoval, A. S. Avaliani and N. A. Gasviani, Soobsch. AN Gruz. SSR, 72, 105 (1973). V. I. Shapoval, V. F. Grischenko and L. I. Zarubitskaya, Ukr. Khim. Zhurn., 39, 867 (1973). V. I. Shapoval, V. F. Grischenko and L. I. Zarubitskaya, Ukr. Khim. Zhurn., 38, 1088 (1972). V. I. Shapoval, Yu. K. Delimarsky and V. F. Grischenko, in Ionic melts, Naukova Dumka, Kiev. 1974, pp. 222-241. O. G. Tsiklauri and N. A. Gasviani, In Materials of Conference of Young Scientists. Inst. of chemistry and electrochemistry of Academy of Sciences of Georgian SSR, Tbilisi, 1976, p. 63. V. I. Shapoval,O. G. Tsiklauri and N. A. Gasviani, Soobsch. AN Gruz. SSR, 88, 609 (1977). Kh. B. Kushkhov and V. I. Shapoval, X All-Union Conference on Physical Chemistry and Electrochemistry of Ionic Melts and Solid Electrolytes, Ekaterinbourg, October 27-29,1992, Ekaterinbourg, 1992, p. 3. R. Combes and B. Tremillon, J. Electroanal. Chem., 83, 297 (1977). M. Tazika, S. Mizoe,M. Nagano and A. Kato, Denki kagaku., 46, 37 (1978). R. Combes, F. De Andrade and L. Carvalho, C. r. Acad. Sci., C285, 137 (1977). V. L. Cherginets and V. V. Banik, Rasplavy, N6, 92 (1990). V. L. Cherginets and V. V. Banik, Rasplavy, N2, 118 (1991). T. P. Boyarchuk, E. G. Khailova and V. L. Cherginets, Ukr. Khim. Zhurn., 58, 758 (1992). V. L. Cherginets and V. V. Banik, Zhurn. Fiz. Khim., 68, 145 (1994). V. L. Cherginets, Potentiometric studies of acidic properties of niobium (V) and germanium (IV) oxides in chloride melts, Kharkov,1991;Dep in ONIITEKhim (Cherkassy),N 241 (1991). Yu. F. Rybkin, V. V. Banik, In Methods of obtaining and investigations of single crystals and scintillators, Inst. Single Crystals,Kharkov,1980, p. 121-125 (1980). Yu. F. Rybkin, V. V. Banik, In Single crystals and engineering, Inst. Single Crystals,Kharkov, 1974, p. 111-114. N. V. Hanf and M. J. Sole, Trans. Faraday Soc., 66, 3065 (1970). Y. F. Rybkin and Y. A. Nesterenko, Zhurn. Fiz. Khim., 50, 781 (1976). R. Combes, J. Vedel and B. Tremillon, Electrochim. Acta, 20, 191 (1975). V. L. Cherginets and V. V. Banik, Rasplavy, (4), 98 (1991). M. V. Smirnov, I. V. Korzun and V. A. Oleynikova, Electrochim. Acta, 33, 781 (1988). C. Butler, J. R. Russel, R. B. Quincy, J. Chem. Phys., 45, 968 (1966). F. Rosenberger, in Ultrapurity methods and Techniques, Dekker Inc, New York, 1972, pp. 3-70. D. Ecklin, Helv. Chim. Acta, 50, 1107 (1967). J. M. Peech, D. A. Bower, R. O. Rohl, J. Appl. Phys., 38, 2166 (1967). U. Gross, Mat. Res. Bull., 5, 117 (1970). M. Lebl and J. Trnka, Z. Phys.,186, 128 (1965). O. V. Demirskaya and Y. A. Nesterenko, In Physics and Chemistry of Crystals, Inst. Single Crystals, Kharkov, 1977, pp. 155-159. Y. F. Rybkin and O. V. Demirskaya, In Single Crystals and Engineering, Inst. Single Crystals, Kharkov, 1974, No. 1(10), pp. 115-118. R. C. Pastor and A. C. Pastor, Mat. Res. Bull., 10, 117 (1975). R. C. Pastor and A. C. Pastor, Mat. Res. Bull., 11, 1043 (1976). A. I. Agulyanskii and P. T. Stangrit, Zhurn. Prikl. Khim., 10, 1201 (1977). V. V. Banik and V. L. Cherginets, VII All-Union Conf. Chem. and Technology of Rare Alkali Elements, Apatity, June 5-10, 1988, Cola Division of Acad. Sci. of USSR, Apatity,1988, p. 116. V. V. Banik, N. I. Davidenko and Y. A. Nesterenko, In Physics and Chemistry of Solids, Inst. Single Crystals, Kharkov, 1983, No. 10, pp. 139-141. J. Ekstein, U. Gross and G. Rubinova, Krist. Tech., 3, 583 (1968).

638

Victor Cherginets

121 N. N. Smirnov and V. R. Lyubinskii, In Single Crystals and Engineering, Inst. Single Crystals, Kharkov, 1971, No. 5, pp. 95-101. 122 O. B. Babushkina and S. V. Volkov, XIV Ukraine Conf. Inorg. Chem., Kiev, September 10-12, 1996, Kiev National University, Kiev, 1996, p. 33.

11

Electronic and Electrical Effects of Solvents 11.1 THEORETICAL TREATMENT OF SOLVENT EFFECTS ON ELECTRONIC AND VIBRATIONAL SPECTRA OF COMPOUNDS IN CONDENSED MEDIA Mati Karelson Department of Chemistry, University of Tartu, Tartu, Estonia

11.1.1 INTRODUCTION The electronic-vibrational spectra of molecules can be substantially influenced by the surrounding condensed medium. The resultant effects arise from a variety of intermolecular interactions between the chromophoric solute and the solvent molecules in such media. Experimentally, these effects can be observed as • the shifts of the spectral maxima (solvatochromic shifts) • the change in the intensity of the spectral line or band • the change of the shape and width of the spectral band Each of those, so-called solvent effects can be described theoretically using different model approaches. The solvatochromic spectral shifts are expected to arise from the difference in the solvation of the ground and the excited states of the molecule. As a result of the spectroscopic excitation, the charge distribution of the molecule changes and thus the interaction will be different in the ground state and in the excited state of the molecule. The direction and size of the respective spectral shift depends directly on the difference in the solvation energy of the molecule in those two states. The larger solvation energy of the ground state (S0), as compared to that of the excited state (S1), results in the negative solvatochromic shift (blue shift) of the spectral maximum (cf. Figure 11.1.1a). Alternatively, the stronger solvation of the excited state, as compared to the solvation of the ground state, leads to the decrease of the excitation energy and is reflected by the positive solvatochromic shift (red shift) in the spectrum of the compound (Figure 11.1.1b). In each case, the direction and the size of the shift depends on the nature and electronic structure of the ground and excited state. For example, in the case of the n → π * transition in acetone (Scheme 11.1.1) an electron from the n-orbital (lone pair) is transferred to the Scheme 11.1.1. antibonding π* orbital localized on the C=O double bond. In

640

Mati Karelson

Figure 11.1.1. The origin of two types of solvatochromic shifts in the spectra of chromophoric compounds.

the ground state, the acetone molecule has a significant dipole moment (2.7 D) arising from the polarity of the C=O bond. Because of the difference in the electronegativity of bonded atoms, the electron distribution is shifted towards the oxygen that could be characterized by a negative partial charge on this atom. However, in the excited state the electron cloud is shifted from the oxygen atom to the bond and, correspondingly, Figure 11.1.2. The schematic representation of the the dipole moment of the molecule is subchange of the solvation energy of the ground (S0) and ex- stantially reduced. In result, the interaction cited state (S1) of acetone moving from a non-polar sol- of the dipole of the solute (acetone) with the vent (1) to a polar non-hydrogen bonded solvent (2) to a surrounding medium in more polar solvents polar hydrogen-bonded solvent (3). is larger in the ground state as compared to the excited state (Figure 11.1.2). Thus, the energy of the ground state is lowered more by the electrostatic solvation than the energy of the excited state. Consequently, the excitation energy increases and the respective spectral maximum is shifted towards the blue end of spectrum (negative solvatochromic shift). In the hydrogen-bonding solvents, the ground state of acetone is ad-

11.1 Theoretical treatment of solvent effects

641

ditionally stabilized by the hydrogen bonding of the oxygen lone pair by the solvent that leads to further increase of the excitation energy (Figure 11.1.2) and the respective blue shift of the spectral maximum. Notably, the formation of this hydrogen bonding is impossible in the S1(nπ*) excited state of the acetone because of the electron transfer from the oxygen lone pair to the antibonding π* orbital. In many cases, the dipole moment increases in the excited state. For instance, in the nitrobenzene (Scheme 11.1.2) the π → π* transition leads to the Scheme 11.1.2. substantial redistribution of the electronic charge reflected by the shift of negative charge on the nitro-group. The real excited state is given by a combination of the La and Lb states. Nevertheless, the dipole moment of nitrobenzene is substantially increased in the first excited state (S1). Because of the substantial charge redistribution in such excited states, they are often called the charge-transfer (CT) states. The much larger dipole of the nitrobenzene in the S1 (ππ∗ ) state is additionally stabilized by a more polar solvents that leads to the decrease in the excitation energy (Figure 3) and to the corresponding solvatochromic red shift of the spectral maximum. In Table 11.1.1, the solvatochromic Figure 11.1.3. The relative energies of the ground (S0) and the first excited state (S1(ππ*)) of nitrobenzene in shifts characterizing various positively and negatively solvatochromic compounds are non-polar (1) and polar (2) solvents. listed. In most cases, the theoretical treatment of the solvatochromic shifts has been based on the calculation of the solvation energies of the chromophoric molecule in the ground and excited states, respectively. Table 11.1.1. The solvatochromic shifts for various positively and negatively solvatochromic compounds Compound

νmax (non-polar solvent), cm-1

νmax (polar solvent), cm-1

∆νmax, cm-1

Ref.

(Scheme 3)

30000 (hexane)

25760 (CF3CH2OH)

+4240

(a)

(Scheme 4)

30170 (hexane)

26140 (water)

+4030

(b)

(Scheme 5)

20640 (hexane)

16860 (water)

+3780

(c)

(Scheme 6)

22620 (hexane)

19920 (DMSO)

+2700

(d)

(Scheme 7)

20730 (hexane)

18410 (methanol)

+2320

(e)

(Scheme 8)

27400 (cyclohexane)

23230 (water)

+4170

(f)

(Scheme 9)

43370 (isooctane)

41220 (water)

+2150

(g)

(Scheme 10)

14600 (toluene)

24100 (water)

-9500

(h)

642

Mati Karelson

Compound



∆νmax, cm-1

Ref.

(Scheme 11)

15480 (tetrahydrofuran)

24450 (water)

-8970

(i)

(Scheme 12)

20410 (chloroform)

22080 (water)

-1670

(j)

(Scheme 13)

20160 (chloroform)

22370 (water)

-2210

(j)

(Scheme 14)

16080 (chloroform)

22170 (water)

-6090

(k)

(Scheme 15)

35870 (CCl4)

37735 (CF3CH2OH)

-1865

(l)

(Scheme 16)

16390 (pyridine)

21280 (water)

-4890

(m)

(Scheme 17)

19560 (chloroform)

22060 (water)

-2500

(n)

S. Spange, D. Keutel, Liebig’s Ann. Chim., 423 (1992). S. Dähne, F. Shob, K.-D. Nolte, R. Radeglia, Ukr. Khim. Zh., 41, 1170 (1975). J.F. Deye, T.A. Berger, A.G. Anderson, Anal. Chem., 62, 615 (1990). D.-M. Shin, K.S. Schanze, D.G. Whitten, J. Am. Chem. Soc., 111, 8494 (1989). E. Buncel, S. Rajagopal, J. Org. Chem., 54, 798 (1989). M.J. Kamlet, E.G. Kayser, J.W. Eastes, W.H. Gilligan, J. Am. Chem. Soc., 95, 5210 (1973). E.M. Kosower, J. Am. Chem. Soc., 80, 3261 (1958). L.S.G. Brooker, A.C. Craig, D.W. Heseltine, P.W. Jenkins, L.L. Lincoln, J. Am. Chem. Soc., 87, 2443 (1965). (i) M.A. Kessler, O.S. Wolfbeis, Chem. Phys. Liquids, 50, 51 (1989). (j) H. Ephardt, P. Fromherz, J. Phys. Chem., 97, 4540 (1993). (k) H.W. Gibson, F.C. Bailey, Tetrahedron, 30, 2043 (1974). (l) G.E. Bennett, K.P. Johnston, J. Phys. Chem., 98, 441 (1994). (m) N.S. Bayliss, E.G. McRae, J. Am. Chem. Soc., 74, 5803 (1952). (n) E.M. Kosower, B.G. Ramsay, J. Am. Chem. Soc., 81, 856 (1959). (a) (b) (c) (d) (e) (f) (g) (h)

Scheme 3

Scheme 4

Scheme 5

Scheme 6

Scheme 7

Scheme 8


Scheme 9

643

Scheme 10

Scheme 11

Scheme 12

Scheme 13

Scheme 14

Scheme 15

Scheme 16

Scheme 17

The solvent-induced broadening of the spectral lines and bands arises primarily from the variation of the local environment of the chromophoric solute molecule in the condensed medium caused by the thermal motion of the surrounding solvent molecules. At any given instant of time, there is a distribution of differently solvated solute molecules, each of which has characteristic transition energy to the excited state. The respective distribution of the transition energies leads to the broadening of the spectral band. It has to be kept in mind, however, that the broadening of spectral lines and bands can be also originated from adjoin-

644

Mati Karelson

ing the rotational and vibrational energy levels in the polyatomic molecule or from the Doppler and natural broadening of spectral lines. Those are more significant in the case of atoms and small molecules. The theoretical assessment of the solvent-induced spectral broadening has thus to rely on a proper statistical treatment of the solvent distribution around the chromophoric solute molecule, both in the ground and in the excited state of the latter. The surrounding solvent can also influence the intensity of the spectral transition (absorption or emission). The intensity of the spectral transition is usually characterized by the oscillator strength f defined as follows  8πm ν  2 f = | M |  3he 2 

[11.1.1]

where m and e are the electron mass and the electron charge, respectively, h is the Planck’s constant, M is the transition moment and ν is the mean absorption wavenumber. Following the last equation, the intensity of the spectrum is proportionally related to transition energy, provided that the transition moment M is independent of the surrounding medium (solvent). This may, however, be not the case. The definition of the transition moment1 M = ∑ ψ 0 q i r i ψ 1*

[11.1.2]

i

includes, apart from the charges (qi) and their position-vectors (ri) in the molecule, the wave function of the molecule in the ground state (ψ0) and in the excited state (ψ *1 ), respectively. Therefore, whenever the solvent affects the wavefunction of the molecule either in the ground state or in the excited state, the intensity of spectral transition is further influenced by the change of the respective transition moment. The analysis of the solvatochromic effects on molecular absorption and emission (fluorescence and phosphorescence) spectra is further complicated by the variation of time scales for the solvent relaxation after the spectral excitation of the solute molecule. The spectral transition is a very fast process that takes place within approximately 10-16 s. Thus, during this short period of time the atomic nuclei do not practically move. The excited state reached by the respective vertical transition is often called the Franck-Condon state (Figure 11.1.4). The lifetime of the fluorescent excited state may be long enough (10-7 - 10-9 s) to allow in addition to the intramolecular nuclear relaxation (10-12 s), also the solvent orientational relaxation. The latter, which is characterized by the relaxation times ranging from 10-10 s up to infinity (in the case of Figure 11.1.4. The Franck-Condon transitions during the solids) may bring up the additional, solexcitation and the de-excitation of the molecule.


645

vent-induced stabilization of the relaxed excited state as compared to the Franck-Condon state. Thus, as a rule, the solvatochromic shifts in the absorption and fluorescence spectra are not equal. The theoretical treatment of the time-dependent effects on molecular spectra in condensed phases is extremely complicated.2 In most cases, it is assumed that only the electronic polarization of the solvent contributes to the solvation energy of the Franck-Condon state (S1 in the case of absorption and S0 in the case of emission). In the case of long-living states, i.e., the ground state and the relaxed excited state, a full relaxation of the solvent is assumed in the field of the solute molecule. The solvation energy of different states at different degrees of relaxation will thus be rather different that may result in rather different dependence of the absorption and emission transition energies on the polarity of the solvent. Some examples of solvatofluorochromical compounds are given in Table 11.1.2. Table 11.1.2. The solvatofluorochromic shifts for various positively and negatively solvatochromic compounds Compound



∆νmax, cm-1

Ref.

(Scheme 18)

24400 (hexane)

16500 (water)

+7900

(a)

(Scheme 19)

21980 (cyclohexane)

18210 (water)

+3770

(b)

(Scheme 20)

~20000 (cyclohexane)

~17000 (CH3CN)

+3000

(c)

(Scheme 21)

23530 (pentane)

22730 (water)

+800

(d)

(Scheme 22)

24150 (hexane)

21850 (methanol)

+2300

(e)

(a) (b) (c) (d) (e)

I.A. Zhmyreva, V.V. Zelinskii, V.P. Kolobkov, N.D. Krasnitskaya, Dokl. Akad. Nauk SSSR, Ser. Khim., 129, 1089 (1959). M. Maroncelli, G.R. Fleming, J. Chem. Phys., 86, 6221 (1987). A. Safarzadeh-Amini, M. Thompson, U.J. Krall, J. Photochem. Photobiol., Part A., 49, 151 (1989). M.S.A. Abdel-Mottaleb, F.M. El-Zawawi, M.S. Antonious, M.M. Abo-Aly, M. El-Feky, J. Photochem. Photobiol., Part A., 46, 99 (1989). J. Catalán, C. Díaz, V. López, P. Pérez, J. Phys. Chem., 100, 18392 (1996).

Scheme 18

Scheme 20

Scheme 19

646

Mati Karelson

Scheme 21

Scheme 22

All theoretical treatments of solvatochromic shifts proceed from modelling the solvational interactions in the liquids and solutions. Theoretically, the interaction potential between a solute molecule and the surrounding solvent molecules Φ is given by the following integral ∞

Φ = C ∫ ϕ (R) g (R) R 2 dR

[11.1.3]

0

where ϕ (R) and g(R) are the pair interaction potential between the solute and the solvent molecule and the solvent radial distribution function around the solute molecule, respectively, and C is a constant depending on the density of the system. The integration in the last formula is carried out over the distance between the solute and the solvent molecule, R. The equation [11.1.3] is derived proceeding from the assumption that the intermolecular forces in the condensed medium are additive. This assumption may be, however, violated because of possible three- and many-body interactions between the molecules in the solution. For most of the real systems, the application of Eq. [11.1.3] directly is rather impractical because of the low precision of ϕ (R) and g(R), particularly in the case of many-atomic molecules. Moreover, this equation will be strictly valid only for the solute-solvent systems in thermodynamic equilibrium and thus not applicable for the Franck-Condon states. Thus, almost all theoretical calculations of solvatochromic effects proceed from different physical models describing the intermolecular interactions in liquids and solutions. Traditionally, the solvation energy of a molecule Esolv in a given solvent can be divided into the following terms3 E solv = E cav + E disp + E elst + E H − bond

[11.1.4]

each of which corresponds to a certain type of intermolecular interaction in the condensed media. Thus, Ecav denotes the energy of the cavity formation for the solute in the solvent, Edisp is the dispersion energy and Eelst the electrostatic energy of the interaction of the solute with the surrounding solvent molecules. The term EH-bond accounts for the energy of the hydrogen bond formation between the solute and solvent molecules. The value of each of the above terms will change as a result of the Franck-Condon excitation of the solute molecule. First, the size of the molecule increases, as a rule, during the excitation. However, as the excitation process is practically instantaneous, the position and orientation of the solvent molecules in the solvation sheath of the chromophoric solute will not change. This means that


647

the average distance between the surface of the solute and the solvent molecules will decrease in the Franck-Condon excited state of the former that normally causes the enhanced solute-solvent repulsion in that state. At the same time, the dispersion energy that stabilizes the solute-solvent system will also increase in the absolute value, but to the opposite direction. In consequence, both effects may cancel each other and the net effect will be close to zero. For the polar solutes, both in the ground and in the excited state, the electrostatic solvation energy is therefore often considered as the most important term in Eq. [11.1.4]. During the excitation or de-excitation of the molecule, the molecular electronic wavefunction and the electron distribution may change significantly. In result, substantial differences are expected in the electrostatic and dispersion solvation energies of the ground and the excited state, respectively. In addition, the hydrogen bonding between the solute and solvent molecules may be affected by the excitation of the solute molecule that will be reflected as another contribution to the difference in the solvation energy of solute in the ground and in the excited state, respectively. In the following, we proceed with the systematic presentation of the theoretical methods developed for the description of the solvatochromic effects on molecular electronic and vibrational spectra in condensed disordered media (liquids, solutions, glasses etc.). 11.1.2 THEORETICAL TREATMENT OF SOLVENT CAVITY EFFECTS ON ELECTRONIC-VIBRATIONAL SPECTRA OF MOLECULES As described above, the change (increase) in the size of the molecule during the excitation will result in increased van-der-Waals repulsion between the electron clouds of the chromophoric solute and the solvent molecules. Alternatively, the size of the molecule is expected to shrink as a result of the de-excitation of the molecule back to the ground state. In such case, the repulsion between the solute and solvent molecules will be reduced correspondingly. The respective energetic effect may be modeled as the difference in the cavity formation energies for the solute molecule in two states. The dependence of the cavity formation energy on the cavity size has been derived using several different model concepts. The simplest approach is based on the concept of microscopic surface tension on the boundary between the solute cavity and the solvent. Within this approach, the free energy of cavity formation is assumed simply proportional to the surface of the solute cavity, SM: ∆G cav = σS M

[11.1.5]

where σ is the surface tension of the solvent. This formula has been applied for the evaluation of the free energy of transfer of electroneutral solutes between different solvents.4 It has been extended to account for the size of the solvent molecule as follows: ∆G cav = σS M − RT ln(1 − Vs ns )

[11.1.6]

where VS is the intrinsic volume of a solvent molecule and nS is the number density of the solvent. In order to account for the chemical individuality of constituent atoms, it has been suggested to use different surface tension values σ i for different atomic types in the solute molecule.5 Thus, ∆G cav = C + ∑ σ i Ai i

[11.1.7]

648

Mati Karelson

where Ai are the solvent-accessible surface areas of different atoms in the solute molecule and C is an empirically adjustable parameter. The quality of the description of experimental data has been, however, not significantly improved by the introduction of individual atomic surface tension characteristics. Another theoretical approach for the calculation of the free energy of cavity formation proceeds from the theory of microscopic curved surfaces. According to this theory,6 V ∆G cav = ksg  S  VM

  σS 

[11.1.8]

where S is the area of the cavity and k gs (VS / VM ) is a correction factor, characteristic of a given solvent and depending on the ratio of molecular volumes of the solvent and solute. This factor has been approximated by the following formula V ksg  S  VM

 V  = 1 +  S   VM

  

2/3

[k

g s

(1) − 1]

[11.1.9]

where k gs (1) is estimated from the solubility of a given solute in a given solvent. The main deficiency of this approach is connected with the introduction of additional empirical information, often not readily available. The free energy of cavity formation has been also estimated from the data on isothermal compressibility, βT, as follows7 ∆G cav =

Vcav +C βT

[11.1.10]

where Vcav is the volume of the cavity and C is a constant term. However, the microscopic isothermal compressibility of water, calculated from the slope of Eq. [11.1.10], was found to be about an order higher than the respective experimental value for water (βT(calc) = 23.5 vs. βT(exp) = 3.14). Therefore, the use of the macroscopic surface tension or compressibility of the solvent for the respective microscopic model quantities is questionable. An entropic approach to the calculation of the free energy of cavity formation proceeds from the scaled particle theory (SPT).8,9 The free energy of the formation of a spherical cavity in a solvent, ∆Gcav, can be calculated proceeding within the framework of SPT as follows 2  9  y    aM   3y  a M  3y ∆Gcav = RT 1− ln(1− y ) +  + +        1− y  a S  1− y 2  1− y    a S  

2

  [11.1.11] 

where y=

4πρas2 3

[11.1.12]


649

is the reduced number density of the solvent. In the two last equations, aM and aS denote the intrinsic radii of the solute and solvent molecules, respectively, and ρ is the number density of the solvent. In the case of an ellipsoidal solute cavity, the SPT cavity formation energy has been given by the following equation10 2   αy  a  βy  y  a  ∆Gcav = RT 1− ln(1− y ) +  + + γ       1− y  a S  1− y  1− y    a S  

2

  

[11.1.13]

where α , β and γ denote the geometrical coefficients and a is the characteristic length of the ellipsoid (the major semi-axis). The scaled particle theory has been extended to dilute solutions of arbitrary shaped solutes and has been successfully applied for the calculation of the solvation free energy of hydrocarbons in aqueous solutions.11 For most practical applications that involve the lowest excited states of the molecules, the increase in the cavity size during the excitation of the solute molecule would not be accompanied with a significant energetic effect. However, it may be important to account for the so-called Pauli repulsion between the solute electronic system and the surrounding medium. This interaction will force the solute electrons to stay inside the cavity and not to penetrate into the dielectric continuum (consisted of electrons, too) that surrounds it. The Pauli repulsion has been modeled by the respective model potentials, e.g., by expanding the potential in spherical Gaussian shells as follows:12

[

VPR = ∑ b i exp −β i (r − ro, i ) i

2

]

[11.1.14]

where bi are the weight factors, βi the exponents and r0,i the radii of spherical shell functions. In general, the electrons in the solvent cavity could be treated as confined many-electron systems.13 11.1.3 THEORETICAL TREATMENT OF SOLVENT ELECTROSTATIC POLARIZATION ON ELECTRONIC-VIBRATIONAL SPECTRA OF MOLECULES The origin of the solvatochromic shifts in the electronic spectra is related to the change in the electrostatic and dispersion forces between the solvent and the chromophoric solute molecule in the ground and in the excited state, respectively. The semiclassical approach to the treatment of the respective effects is based on the assumption that the solute and the solvent molecules are sufficiently separated to neglect the overlap between the electron distribution of these two molecular systems. The wave function for the whole system can then be approximated as the product of the wavefunctions of each individual system, i.e., the solute and individual solvent molecules: Ψ = ψ s0(1) ψ s0( 2 ) L ψ s0( n ) ψ (aν )

[11.1.15]

0 0 , etc. are the wavefunctions of the respective solvent molecules in where ψ 0s(1) , ψ s(2) ,L ψ s(n) the ground state and ψ (aν ) is the wavefunction of the solute molecule in the ν-th state. The antisymmetry of the total electronic wavefunction is ignored as the individual molecules are assumed separated enough not to allow the electron exchange. This approximation may not be valid in the case of strong semichemical interactions between the solute and the solvent

650

Mati Karelson

molecules such as the hydrogen bonding or the formation of charge-transfer complexes. In such cases, the system consisting of the central solute molecule and the adjacent solvent molecules has to be treated as a supermolecule. In the absence of strong semichemical interactions between the solute and solvent molecules, the interaction energy between them can be derived using the perturbation theory.14 In the first approximation, the interaction between the nonionic molecules can be reduced to the dipole-dipole interaction between the molecules. The following perturbation operator can describe this interaction $ µ$ µ$ (αv ) Θ ai s ( i ) $ ′ H =∑ Rai3 i

[11.1.16]

where µ$ (αν ) and µ$ s(i) are the dipole moment operators for the solute a in the ν-th state and for the i-th solvent molecule in the ground state, respectively, Rai is the distance between the charge centroids of the interacting molecules and →

→

$ = 1$ − 3 R ai R ai Θ ai

[11.1.17]

is the angular term describing the relative orientation of these two molecules in the space. The subsequent application of the perturbation theory to derive the energy of interaction between a pair of a solute and a set of N solvent molecules gives the following result15 N

µ (aν ) Θ ai µ s ( i )

i =1

Rai3

∆E a = ∑ −

−

(ν ) (ν ) 1 N µ a Θ ai α s Θ ai µ a − ∑ 2 i =1 Rai6

( ν) µ ( 0 p ) Θ µ ( λν)µ ( λν) Θaj µ (s0( ip) ) 1 N N µ s ( i ) Θai α a Θaj µ s ( j ) N − ∑ ∑ ∑ s ( i ) ai a 6 a ∑ ∑ 3 3 Rai Raj Rai 2 i =1 j =1 i =1 p ≠ 0 λ≠ ν

[11.1.18]

with the following notations: µ (aν ) = ψ (aν ) µ$ a ψ (aν )

[11.1.19]

is the dipole moment of the solute in the ν-th state, µ s ( i ) = ψ s0( i ) µ$ s ψ s0( i )

[11.1.20]

is the dipole moment of the i-th solvent molecule in the ground state, α

(ν ) a

= 2∑ λ≠ν

ψ (aν ) µ$ a ψ (aλ ) ψ a( λ ) µ$ a ψ a( ν ) Eν −Eλ

is the polarizability tensor of the solute molecule in the ν-th state,

[11.1.21]


α s = 2∑

p≠0

ψ s0 µ$ s ψ s( p ) ψ s( p ) µ$ s ψ s0 E 0s − E ps

651

[11.1.22]

is the polarizability tensor of the solvent molecule in the ground state, respectively, and µ s( 0( ip) ) = ψ s0( i ) µ$ a ψ s( p( i))

[11.1.23]

µ (aλν ) = ψ (aλ ) µ$ a ψ (aν )

[11.1.24]

and

are the transition dipoles between the two states (0 → p) in the solvent and in the solute (λ → ν) molecules, respectively. In the last formulae, Eν and Eλ denote the energy of the solute molecule in the respective (ν-th and λ-th) states, and Eps and E0s - the energy of a solvent molecule in the p-th and in the ground state, respectively. The first term in equation [11.1.18] represents therefore the electrostatic interaction of the unperturbed charge distribution of the two molecules, given as the interaction between the respective permanent point dipoles. The second term in this equation corresponds to the interaction of the permanent dipole of the solute with the dipole induced in the solvent whereas the third term reflects the interaction of the permanent dipole of the solvent with the induced dipole of the solute. The last term represents the second-order interaction of both molecules in excited states and quantifies thus effectively the dispersion interaction in the solute-solvent system. The equation [11.1.18] refers, of course, to a single fixed configuration of the solute and the solvent molecules. In order to find the effective interaction energy in the liquid medium, an appropriate statistical averaging over all configurations has to be carried out. In most practical applications, this procedure is very complicated and thus the semiclassical continuum approaches are employed to describe the solvent. The description of the electrostatic interactions between the solute and the solvent has been based on the Onsager’s reaction field model. According to this model, the energy of electrostatic interaction between an arbitrary charge distribution inside the solute molecule and the surrounding polarizable dielectric medium is given by the following equation16 E el =

1 Es E o (ε − 1)dV 8π V∫

[11.1.25]

where Eo is the electrostatic field of the charges in the molecule in vacuo and Es is the modified field in the presence of dielectric medium. Notably, within the formalism of the last [11.1.25], the dielectric constant ε of the medium is still a function of the space coordinates, i.e., both the interior of the molecule and the surrounding medium are treated by the same equation. However, the integral in the last equation cannot be found analytically and even the numerical integration over the space presents a difficult mathematical task. Therefore, the electrostatic equation is usually simplified by the application of the Gauss divergence theorem. According to this theorem, the volume integral in [11.1.25] is transformed into a surface integral over some boundary

652

Mati Karelson

E el =

r ε −1 ΦS E o ndS ∫ 8π S

[11.1.26]

r where S is the boundary surface, n the outward normal unit vector on S and the reaction potential Φ s is defined as follows: Es = -grad ΦS. Depending on the shape of solute molecular cavity, different approaches have been applied for the calculation of the electrostatic solvation energies of compounds in liquids. Within the classical reaction field theory of Kirkwood and Onsager,17,18 the solute molecule is represented by a set of point charges fixed inside of sphere of a radius a 0 and the electrostatic equation [11.1.26] is solved by applying the appropriate boundary conditions inside and outside the sphere. It is also assumed that the dielectric constant inside the cavity (sphere) is equal to unity (vacuum) and outside the cavity has a constant value, corresponding to the macroscopic dielectric constant of the medium studied. In that case, the energy of the electrostatic interaction between the solute charge distribution and the surrounding dielectric medium is given by the following infinite expansion E el

l l ∞ (l + 1)(1 − ε)  r r 1 i j = ∑ ei e j ∑   2 l+1 Pl cosθ ij 2 i, j l = 0  ε(l + 1) + 1  a 0

(

)

[11.1.27]

where ei and ej are the charges inside the sphere at positions ri and rj, respectively, and θij is the angle at the center of the sphere between the vectors ri and rj. In the last equation, the summation proceeds over all charged particles (nuclei and electrons) of the solute and P1(cosθij) are the Legendre polynomials of l-th order. By expressing the Legendre polynomials as the products of the respective spherical harmonics of order m (-l ≤ m ≤ l), equation [11.1.27] can be rewritten as E el = −

1 ∞ l ∑ ∑ Rlm M lm 2 l = 0 m = −1

[11.1.28]

where Rlm = f l M lm

[11.1.29]

and fl =

(l + 1)(ε − 1) 1 (l + 1)ε + 1 a 02 l+1

[11.1.30]

In these equations, M ml and R ml represent the electrical momentum and the respective reaction field component. The first term (l = 0) in the expansion [11.1.27] gives the interaction of the excess (ionic) charge of the solute with the respective reaction field created in the dielectric medium (Born term) E Born =

1− ε Q 2 2ε a 0

[11.1.31]


653

where Q is the numerical value of the ionic charge. The next term (l = 0) corresponds to the total dipole interaction with the corresponding reaction field (Onsager dipolar term) E Onsager

r (1 − ε) µ 2 = ( 2ε + 1) a 03

[11.1.32]

r where µ is the dipole moment of the solute. In most applications, only these two terms that are the largest by size are considered in the calculation of the electrostatic interaction energy. However, depending on system studied, the interaction of higher electrical moments with the corresponding reaction field may become also significant and the terms corresponding to higher moments of order 2l (quadruple, octuple, hexadecuple, etc.) should be taken into account. In many cases, the shape of the solute molecule may be very different from the sphere and therefore, it is necessary to develop the methods of calculation of the electrostatic solvation energy for more complex cavities. In the case of the ellipsoidal cavity with main semiaxes a, b, and c, the analytical formulas are still available for the calculation of the charge and dipolar terms of the electrostatic interaction with the reaction field. The charge term is simply ell E Born =

(1 − ε) Q 2 2ε abc

[11.1.33]

whereas in the respective dipolar term19 r ell E Onsager = Rµ

[11.1.34]

the reaction field R is presented using a special tensor as follows  3Aa (1− Aa )( 1− ε)µ a   abc[ε + ( 1− ε)Aa ]  R = 0   0  

0 3Ab (1− Ab )( 1− ε)µ b abc[ε + ( 1− ε)Ab ] 0

   r µ 0  3Ac (1− Ac )( 1− ε)µ c   abc[ε + ( 1− ε)Ac ]  0

[11.1.35]

where Aa, Ab and Ac are the standard ellipsoidal shape factor integrals, and µa, µb and µc are the dipole moment components along the main semiaxes of the ellipsoid. Several methods have been developed to define the semiaxes of the ellipsoidal cavity. For instance, these have been taken collinear with the axes of the solute dipole polarizability tensor, and their lengths proportional to the respective eigenvalues.20 Another definition proceeds from the inertia tensor of the van-der-Waals solid, i.e., a solid or uniform density composed of interlocking van-der-Waals spheres.21 Also, the ellipsoidal surface has been defined in terms of the best fitting of a given molecular electrostatic isopotential surface.22 The above-discussed theoretical formulation of the electrostatic solute-solvent interaction is applicable for the fixed charge distribution inside the solute molecule. However, the solvent reaction field may cause a redistribution of the charge inside the solute. The

654

Mati Karelson

magnitude of this redistribution depends on both the dielectric constant of the solvent and the polarizability of the solute molecule. Within the approximations of the spherical solute cavity and the point dipole interactions between the solute and solvent, the dynamically changed Onsager reaction field can be expressed by the following formula14 Rl =

µ (aν ) + α (aν ) Rl 2(ε − 1) a 03 (2ε + 1)

[11.1.36]

Notably, the use of the macroscopic dielectric constant ε = ε0 in the last formula is justified only when the lifetime of the solute molecule in a given (ν-th) state is much longer than the rotational-vibrational relaxation time of the solvent at given temperature. This is not a valid assumption in the case of the Franck-Condon states, which have the lifetime much shorter than the rotational-vibrational relaxation time of the solvent. Therefore, the solvent is only partially relaxed for these states and the corresponding reaction field is characterized by the dielectric constant at infinite frequency of external electric field, ε ∞ . By inserting the expression for the reaction field [11.1.36] into the equation [11.1.18] and assuming that the static polarizability of the solute molecule is approximately equal to the one third of the cube of Onsager’s cavity radius α a( ν ) ≈

a 03 3

[11.1.37]

the following semiclassical equation can be obtained for the solvation energy of the ν-th (Franck-Condon) state of the solute molecule14  ε − 1 ε∞ − 1   2µ 0aµ (aν) 2µ 0aµ (aν)   + Es = −  0 − 3 a30  ε0 + 2 ε∞ + 2   a 0

( )

( ν)  ε0 − 1   2 µ a   −  a30  ε0 + 2  

(

)

2

 ε∞ − 1    +  ε∞ + 2 

2

  3  (2ε0 + 1) 2 ( ε0 − ε∞ ) 2 kT  2ε − 2  2  µ (aλν) 0    + + +    ∑ (2ε0 − ε∞ )ε0 3a30  2ε0 + 1  λ≠ ν Eλ − Eν   2ε0 + 1  ε − 1 2  3 +  ∞  ε∞ + 2  a 0

∑ (µ ) λ≠ ν

( λν ) a

2

 E − Eλ  1− ν   E −E  ps  0s 

[11.1.38]

The solvatochromic shift due to the difference in the electrostatic solvation energy of the ground state and the excited state of the solute, respectively, is thus given as follows: ∆E s =

2   ε 0 − 1 ε ∞ − 1   2ε 0 + 1   −  a 03   ε 0 + 2 ε ∞ + 2   ε 0 + 2 

  3 +     2ε0 +

[(µ

0 a

)

2

 ε − 1  − µ a0µ a(ν ) +  ∞  ε∞ + 2 

]

2 2  (2ε0 + 1) (ε0 − ε∞ ) kT  (µ (a0 λ ) ) (µ a(νλ) )2   +  −   ∑ ∑ 1 (2ε0 − ε∞ )ε0 3a03  λ≠0 E 0 − E λ λ≠ν E ν − E λ  

[(µ

0 a



) − (µ ) ] 2

(ν ) a

2




 2ε − 2   +  0  ε0 + 2 

2

 (µ a(ν ) )2  6  a 0

(µ ) ∑E − E (νλ ) a

λ≠ν

ν

2

− λ

655

(µ ) (0 ) a 6 0

a

 ε − 1 2  E − E0 2  3  ∑(µ (a0 λ ) )  1− λ +  ∞  + ε 2 a E 0 ≠ λ 0 s − E ps  ∞  0  

2

(µ ) ∑E − E (0 λ ) a

λ≠0

0

 + λ  2

  E − Eν  − ∑(µ (aνλ ) )2  1− λ   λ≠ν E ps − E 0 s  

    

[11.1.39]

The last expression represents the solvent effect on the transition energy of the 0-0 band of the solute molecule. McRae15,23 has given a different derivation of the electrostatic solvation energy based on semiclassical reaction field approach. The final result is however similar to the above equation for the solvatochromic shift in the electronic transition:15 ∆E s =

+

1 a03

2   ε 0 − 1 ε ∞ − 1   2ε 0 + 1   −  a 03   ε 0 + 2 ε ∞ + 2   ε 0 + 2 

[(µ

0 a

)

2

 − µ a0µ (aν )  + 

]

 ε∞ − 1  2 2 1 1 2 2   (µ a(ν ) ) − (µ 0a ) − α a(ν ) E a(ν ) + α 0a E a0 + Da(ν ) − Da0 + ε 2 2 2  ∞ 

[

]

[11.1.40]

where E0a and E(aν ) are the solvent fields due to the permanent dipole moments of the solvent molecules applying to the ground state and to the excited state of the solute molecule, respectively. The terms D0a and D(aν ) denote the solute-solvent intermolecular dispersion energies in the corresponding states. Abe24 has developed an alternative semiclassical theory of the solvent effects on electronic spectra. This theory is based on the averaging of the intermolecular interaction energy over all solute-solvent configurations within the approximation of pair interactions. The theory involves the dipole moments and polarizabilities of the solute molecule and takes into account the temperature dependence arising from the Boltzmann factor. In all above-listed theoretical approaches, the response of the solute charge distribution to the solvent field is expressed by using the static polarizability of the solute molecule. However, it would be plausible to account for this response directly within the quantum mechanical theoretical framework. The quantum-chemical approaches to the calculation of the solvation effects on the ground and excited states of the molecules in the solution can be classified using two possible ways. First, it can be based on the traditional division of the quantum chemistry into the non-empirical (ab initio) and the semiempirical methods. Within both those classes of methods, the Hartree-Fock method based on the independent particle model and the methods accounting for the static and dynamic electron correlation are usually distinguished. The second way of classification of methods can be based on the differences of the models used for the description of solute-solvent interactions. In general, these interactions can be taken into account in the framework of continuum representation of the solvent or using the resolution of solute-solvent interactions at molecular level. In the following, we first proceed with the review of models used for the solute-solvent interactions, with the subsequent elaboration of the quantum-chemical methodology for the calculation of the solvent effects on spectra. The simplest continuum model is based on the classical Onsager reaction field theory assuming the spherical or ellipsoidal form of cavities for the solute molecules in dielectric

656

Mati Karelson

media. The respective interaction energy is accounted for as a perturbation V$ (a 0 , ε ) of the $ 0. Hamiltonian of the isolated solute molecule, H $ =H $ 0 + V$ (a , ε) H 0

[11.1.41]

Within the approximation of electrostatic interaction between the solute dipole and the respective reaction field, the perturbation term is simply V$ (a 0 , ε) = Γµ$ 2a

[11.1.42]

where Γ=

2(1 − ε)

[11.1.43]

(2ε + 1)a 03

In the case of ellipsoidal cavities, the last coefficient has to be substituted by the tensor given in equation [11.1.35]. A self-consistent reaction field method (SCRF) has been developed at the level of Hartree-Fock theory to solve the respective Schrödinger equation25 $ = EΨ HΨ

[11.1.44]

Proceeding from the classical expression for the electrostatic solvation energy of a solute molecule in a dielectric medium in the dipole-dipole interaction approximation, the total energy of the solute is presented as follows26 E = Eo −

(

r r 1 Γ ψ µ$ ψ ψ µ$ ψ + 2µ nuc ψ µ$ ψ + µ 2nuc 2 ∧

)

[11.1.45]

∧

where Eo = , H0 is the Hamiltonian for the reaction field unperturbed solute molecule and ψ is the molecular electronic wave function. From the last equation, one can construct the variational functional L = Eo −

(

)

r r 1 Γ ψ µ$ ψ ψ µ$ ψ + 2µ nuc ψ µ$ ψ + µ 2nuc − W ( ψ | ψ − 1) [11.1.46] 2

where W is the Lagrange multiplier ensuring the normalization of the variational wave function. The variation of the last equation with respect to the parameters of the wave function yields

(

r δL = δE o − Γ δψ µ$ ψ ψ µ$ ψ + µ nuc ψ µ$ ψ

) −Wδ( ψ | ψ )

$ 0 ψ − Γ δψ µ$ ψ µr − W δψ | ψ + c. c. = 0 = δψ H tot

[11.1.47]

r where µ$ tot = µ$ nuc + is the total dipole moment of the solute molecule. The latter is calculated during the SCRF procedure simultaneously with the total energy of the system.


657

From equation [11.1.47], the following Schrödinger equation is obtained for the electronic state |ψ > of the solute molecule $ 0 − Γ ψ µ$ ψ µ$ = W | ψ H tot el

[11.1.48]

W plays the role of the quantum mechanical motif that is directly obtained from the Schrödinger equation as follows $ 0 − Γ ψ µ$ ψ µ$ ψ = ψ H $ 0 ψ − Γ ψ µ$ ψ ψ µ$ ψ W = ψH tot el tot el

[11.1.49]

By adding the part for the interaction of the nuclear component of the solute dipole with the total reaction field and assuming $ 0 ψ ≈ ψ0 H $ 0 ψ0 ψH

[11.1.50]

where ψ 0 is the wavefunction of the solute molecule, unperturbed by the reaction field, one obtains that $0ψ −Γ E el = E rf − E 0 = ψ H

( ψµ$

el

r ψ + µ nuc

)

2

$ 0 ψ 0 ≈ − Γ µr − ψ0 H tot

2

[11.1.51]

The comparison of the last equation with the starting equation [11.1.44] reveals a difference by the factor of two in the final result. Of course, the approximation [11.1.50] brings up a certain error and it has been therefore proposed25,27 to correct the last formula by the addition of the “solvent cost”, i.e., the additional work required to reorganize the solvent due to the electrostatic field of the solute E el = E rf − E 0 +

1 r Γ µ tot 2

2

[11.1.52]

Alternatively, the electrostatic solvation energy can be derived proceeding from the following variational functional27 L = E0 −

(

)

r r2 1 − W ( ψ | ψ − 1) [11.1.53] Γ ψ µ$ ψ ψ µ$ ψ + 2µ nuc ψ µ$ ψ + µ nuc 4

This leads to a Schrödinger equation which eigenvalue is directly related to the total electrostatic (dipolar) solvation energy, Eel,  $0 1 (e )  H − Γ ψ µ$ tot ψ µ$ el  | ψ = E el | ψ 2   E el = E rf( e ) −

1 r r Γµ nuc µ tot − E 0 2

[11.1.54]

[11.1.55]

658

Mati Karelson

For the solution of equation [11.1.54], the molecular wavefunction can be presented as a proper spin-projected antisymmetrized product of molecular (or atomic) orbitals27 |ψ§ = Os A[φ1,L, φn ]

[11.1.56]

Recalling that the dipole moment operator is a one-electron operator (as are all electric moment operators), the following orbital equations are obtained f (k)φi (k ) = ε i φi (k )

[11.1.57]

f (k) = f 0 (k) − Γ ψ µ$ tot ψ µ$ el (k)

[11.1.58]

with

or f (k) = f 0 (k) −

1 Γ ψ µ$ tot ψ µ$ el (k) 2

[11.1.59]

where f0(k) is the usual Fock operator for the isolated molecule, ε i is the molecular orbital energy for |φi > and µ$ el (k) is the electronic part of the dipole moment operator. Both equations are solved iteratively, using the usual SCF procedure and the expectation value of the total dipole moment from the previous SCF cycle. A scheme for the treatment of the solvent effects on the electronic absorption spectra in solution had been proposed in the framework of the electrostatic SCRF model and quantum chemical configuration interaction (CI) method.27 Within this approach, the absorption of the light by chromophoric molecules was considered as an instantaneous process. Therefore, during the photon absorption no change in the solvent orientational polarization was expected. Only the electronic polarization of solvent would respond to the changed electron density of the solute molecule in its excited (Franck-Condon) state. Consequently, the solvent orientation for the excited state remains the same as it was for the ground state, the solvent electronic polarization, however, must reflect the excited state dipole and other electric moments of the molecule. Considering the SCRF Hamiltonian $ =H $ 0 + Γ φµ$ φ µ$ H tot

[11.1.60]

it is possible to write for the state |ψ I ¦ the following expression $ 0 + Γ ψ µ$ ψ µ$ ψ = ψ H $ 0 ψ − Γ ψ µ$ ψ ψ µ$ ψ HII = ψ I H I I I I I 0 0

[11.1.61]

that is the zeroth order estimate of the energy of the state |ψ I ¦. Then, for a single excitation, I → A, the excitation energy is given as follows


659

 2  2 ∆E ia = εa − εi − J ia +  Kia = εa0 − εi0 − J ia +  Kia − 0    0 − Γ ψ 0 µ$ ψ 0 ψ ai µ$ ψ ai − ψ 0 µ$ ψ 0

[

[11.1.62]

]

where Jia and Kia are the respective Coulomb’ and exchange matrix elements and Γ is the reaction field tensor at the dipole level. The terms ε 0i are the eigenvalues of the Fock operator for the k-th electron in the isolated solute molecule. The off-diagonal CI matrix elements are given by $ 0 + Γ ψ µ$ ψ µ$ ψ = ψ H $ 0 ψ − Γ ψ µ$ ψ ψ µ$ ψ HIJ = ψ I H J I J I J 0 0

[11.1.63]

Equations [11.1.60] - [11.1.63] demonstrate that some part of the solvent effect is already included in the ordinary CI treatment when proceeding from the SCRF Fock matrix. It $ 0 ¦ should represent the ground-state dipole moment has to be noticed that the terms ¢ψ 0 |µ|ψ after CI, and therefore, an iterative procedure would be required to obtain a proper solution. However, at the CIS (CI single excitations) level, commonly used for the spectroscopic calculations, this is no concern because of Brillouin’s theorem, which implies that the CI does not change the dipole moment of the molecule. Even at higher levels of excitation in CI, this effect should not be large and might be estimated from the respective perturbation operator.27 There are two approaches to address the instantaneous electronic polarization of the solvent during the excitation of the solute molecule. In the first case, the following correction term has to be added to the CI excitation energy ∆E I =

1 Γ(ε ∞ ) ψ 0 µ$ ψ 0 ψ I µ$ ψ I − ψ I µ$ ψ I  2

2

 

[11.1.64]

where Γ( ε ∞ ) is the reaction field tensor for the optical relative dielectric permittivity of the solvent, ε ∞ . In the last equation, the first term removes the incorrect term arising from the SCRF orbitals and energies in forming the CI matrix whereas the second term adds the response of the electronic polarization of the solvent to the dipole of the excited state. Equation [11.1.64] is first order in electron relaxation. Higher orders can be examined by defining the perturbation X ( I ) = λΓ(ε ∞ )

[ψ

0

µ$ ψ 0 − ψ I µ$ ψ I

]

[11.1.65]

2

which is clearly different for each excited state and would, if pursued, lead to a set of excited states that were nonorthogonal. In principle, these corrections need not to be small. Depending on the Fock operator used (equation [11.1.58] or [11.1.59]), the excitation energy from the ground state |ψ 0 ¦ to the excited state |ψ I ¦ of a solute molecule in a dielectric medium is given as follows $ψ − ψ H $ ψ + 1 Γ ψ µ$ ψ WI A − W0A = ψ I H I 0 0 0 0 2

[ ψ µ$ ψ I

I

− ψ 0 µ$ ψ 0

]−

660

Mati Karelson

1 − Γ(ε ∞ ) ψ I µ$ ψ I 2

[ψ

I

µ$ ψ I − ψ 0 µ$ ψ 0

]

[11.1.66]

in the first case, and $ψ − ψ H $ ψ − 1 Γ (ε ) ψ µ$ ψ W I B − W 0B = ψ I H I 0 0 ∞ I Ι 2

[ ψ µ$ ψ I

I

− ψ 0 µ$ ψ 0

]

[11.1.67]

in the second case. The last two equations are first order in electron polarization of the solvent. The second approach to the calculation of spectra in solutions is based on the assumption that the ground and excited states are intimately coupled in an instantaneous absorption process.28,29 In this model, the solute ground state electron distribution responds to the electron distribution in the excited state through the instantaneous polarization of the solvent. In such a case, the energy of the absorbing (ground) state is shifted by the following amount 1 1 1  Γ (ε∞ ) ψ 0 µ$ ψ 0 − ψ 0 µ$ ψ 0 − Γ (ε∞ ) ψ 0 µ$ ψ 0  ψ I µ$ ψ i − ψ 0 µ$ ψ 0  2 2 2 

[11.1.68]

In the last equation, the first term removes the first order in the electron polarization part of the dielectric relaxation included in the SCRF of the ground state, and the second term adds back the appropriate interaction of the ground state with the “mean” reaction field, created by the excited state, |ψ I ¦ . This leads to the following equation for the excitation energy $ψ − ψ H $ ψ + 1 Γ ψ µ$ ψ WI A − W0A = ψ I H I 0 0 0 0 2

1 − Γ(ε ∞ ) ψ I µ$ ψ I 4

[ψ

I

µ$ ψ I − ψ 0 µ$ ψ 0

]

[ ψ µ$ ψ I

I

− ψ 0 µ$ ψ 0

2

]+

[11.1.69]

for the Fock operator [11.1.58] and

[

$ψ − ψ H $ ψ + 1 Γ (ε ) ψ µ$ ψ − ψ µ$ ψ W I B − W 0B = ψ I H I 0 0 ∞ I I 0 0 4

]

2

[11.1.70]

for the Fock operator [11.1.59]. However, it should be noticed that all four equations for spectral transition energies [[11.1.66], [11.1.67], [11.1.69] and [11.1.70]] yield very similar results when π → π * and n → π* transition solvatochromic shifts have been compared between the nonpolar and polar solvents.27 In Table 11.1.3, the results of the INDO/S (ZINDO)30,31 SCRF CIS calculated spectroscopic transition energies are given for some solvatochromic dyes. The relative shifts due to the solvent are reproduced theoretically in most cases. Even the absolute values of the spectroscopic transition energies are in satisfactory agreement with the respective experimental values, which demonstrates the applicability of the spectroscopic INDO/S parameterization for the spectra in solutions.


661

Table 11.1.3. INDO/S SCRF CI calculated and experimental spectroscopic transition energies of some dyes in different solvents27 Molecule

Solvent

νcalc, cm-1

νexp, cm-1

(Scheme 23)

Gas phase Cyclohexane Water

29,700 26,300 22,500

27,400 23,300

(Scheme 24)

Gas phase n-Hexane Water

36,900 34,800 31,100

30,200 26,100

(Scheme 25)

Gas phase Chloroform Water

20,200 21,800 24,600

19,600 22,100

Scheme 23

Scheme 24

Scheme 25

The INDO/S SCRF CI method has been also successfully applied for the prediction of the solvatochromic shifts in various nitro-substituted porphyrins.32 The SCRF methodology has been employed also for the prediction of the solvatochromic shifts on emission spectra.33,34 A satisfactory agreement was obtained between the calculated and experimental fluorescence energies of p-N,N-dimethylaminobenzonitrile in different solvents. Finally, the solvent-induced shifts in the vibrational spectra of molecules have been also calculated using the SCRF theory.35 The SCRF approach has been also implemented for the treatment of solute-continuum solvent systems at the ab initio Hartree-Fock level of theory.36,37 In addition, a general SCRF (GSCRF) approach has been proposed to account for the interaction of the solvent reaction field with the arbitrary charge distribution of the solute molecule. According to this theory,38,39 the effective Hamiltonian of the solute in the solvent has the following form $ =H $ 0 + drΩ (r) V 0 (r) + dr ′G(r,r ′)Ω (r ′) H s s m s ∫ s ∫





[11.1.71]

where Ω s ( r ) is the solute charge density operator given by Ω s (r) = −∑ δ(r − ri ) + ∑ Z a δ(r − R a ) i

a

[11.1.72]

662

Mati Karelson

where ri stands for the i-th electron position vector operator, Ra is the position vector of the a-th nucleus with the charge Za in the solute and δ(r) is the Dirac’s delta function. The first term in square brackets in Eq. [11.1.71], Vm0 ( r ), represents the electrostatic potential created by the solvent in the absence of the solute and the second, integral term corresponds to the reaction potential response function of the polarizable solvent. Together these terms produce the reaction field potential applying to the solvent molecule in the polarizable dielectric medium. Notably, a principal part of the Hamiltonian [11.1.71] is the solute charge density that can be represented using different approximations of which the multipolar expansion has been mostly applied. By using the distributed multipole model, it is possible to obtain the GSCRF equations for the molecules of a complex shape. However, it has been mentioned that the use of multipole expansions of the solvent electrostatic and reaction potentials in Eq. [11.1.71] may cause this Hamiltonian to become unbound and special damping procedures have been invented to overcome this difficulty. The GSCRF total energy of the solute is given by the following equation $ 0 ψ + ψ drΩ ( r )V 0 ( r ) ψ + EGSCRF = ψ H ∫ s m

1 ψ dr Ω s ( r ) ψ 2 ∫

∫ dr ′G ( r, r ′)Ω ( r ′) [11.1.73] s

This energy expression can be used to build up the respective variational functional to get the molecular orbitals [above]. A crucial step in the general self-consistent reaction field procedure is the estimation of the solvent charge density needed to obtain the response function G(r,r') and the reaction potential. The use of Monte Carlo or molecular dynamics simulations of the system consisting the solute and surrounding solvent molecules has been proposed to find the respective solvent static and polarization densities. Several methods have been developed to account for the solute cavities of arbitrary shape in the solution. The polarizable continuum model (PCM) is based on the numerical integration of the relevant electrostatic equations describing the electrostatic interaction between the molecular charge distribution and the charge created on the boundary surface between the solute molecule and surrounding dielectric continuum.40-45 Within this method, the solute cavity is usually constructed from the overlapping van der Waals spheres of constituent atoms in the solute molecule and the solvent reaction field arising from the solute charge distribution is calculated numerically. Alternatively, the cavity can be defined as constructed from the electron isodensity surface around the solute molecule (IPCM).46 According to the classical electrostatics, the electrostatic potential at any point in the space can be described in terms of the apparent charge distribution, σ, on the cavity surface. It consists of two terms Φs = Φ M + Φ σ

[11.1.74]

the first of which (Φ M ) corresponds to the electrostatic potential created by the charge distribution of the solute and the second (Φ σ ) is due to the reaction potential by the solvent. The latter is directly connected with the apparent charge distribution on the surface of the cavity as follows: Φ σ (r) = ∫

Σ

σ(s) r −s

d 2s

[11.1.75]


663

where Σ is the cavity surface and s vector defines a point on Σ. The Σ surface is usually divided into appropriate number of triangular small areas (tesserea), each of which has an area ∆Sk and contains the charge qk in some internal point sk. Thus, according to this, so-called boundary element method, the reaction potential is found as the following sum over all tesserea Φ σ (r) = ∑ k

qk r − sk

[11.1.76]

with q k = ∆S k σ(s k )

[11.1.77]

In the application of the boundary element method, it is crucial to select appropriate boundary surface for the solute cavity and to proceed as accurate as possible tessellation (triangulation) of this surface. For instance, it has been proposed that in the case of the cavity formation from overlapping van-der-Waals spheres, the atomic van-der-Waals radii should be multiplied by a coefficient equal to 1.2. Other possibilities of the surface definition include the closed envelope obtained by rolling a spherical probe of adequate diameter on the van-der-Waals surface of the solute molecule and the surface obtained from the positions of the center of such spherical probe around the solute. Within the quantum-mechanical theory, the PCM model proceeds from the following Schrödinger equation for a solute molecule in the dielectric continuum

(H$

0

)

+ V$PCM Ψ = EΨ

[11.1.78]

where the reaction field potential is given by equation [11.1.76] as follows qk V$PCM = ∑ k r − sk

[11.1.79]

The charges on the boundary are found from the electrostatic polarization of the dielectric medium on the surface of the cavity due to the potential derived from the charge distribution of the solute and from other (induced) charges on the surface. The induced surface charge is evaluated iteratively at each step of the SCF procedure to solve the Schrödinger equation [11.1.78]. It has been reported that a simultaneous iteration of the surface charge with the Fock procedure reduces substantially the computation time without the loss in the precision of calculations. Also, it has been shown that the expressions which determine the charges qk may be given as a set of linear equations. In the matrix form42 Dq 0f = Ein

[11.1.80]

where D is a square nonsymmetric and nondiagonal matrix with the dimension equal to the number of surface elements, derived from the curvature of the surface and q0f is a column

664

Mati Karelson

vector, containing the unknown surface charges. Ein is also a column vector collecting the effective components of the solute electric field multiplied by the surface elements r [11.1.81] E in0 , k = −∆S k ∇Φ M , in (s k )n k The D matrix depends only on the shape of the cavity and the dielectric constant of the medium. Therefore, when the system of linear equations [11.1.80] has to be solved several times with different Ein, as in the case of the polarizable solute, it may be convenient to work with the single inversion matrix D-1. The PCM model has been implemented for the calculation of the electronic excitation energies of solvated molecules within the quantum-mechanical configuration interaction method.47,48 The respective final expression for the excitation energy from the ground state (0) to the I-th state has the following form ∆W (0 I ) = ∆ECI(0 I ) − +

1 [J20(PI − P0) + P0 T0(PI − P0)] + 2

[11.1.82]

1 [J2∞(PI − P0) + PI T∞(PI − P0)] 2

where PI and P0 are the electronic density matrices of the solute in the excited state and in the ground state, respectively, ∆ECI( 0 I ) = EI − E0 +

[

]

1 2( PI − P0 ) T0 P0 + ( PI − P0 ) J10 + J20 ( PI − P0 ) 2

[11.1.83]

are the diagonal elements of CI matrix, and T = C t W −1 ∂C$

[11.1.84]

$ J1 = CW −1 ∂MZ

[11.1.85]

J 2 = Z t M t W −1 ∂C$

[11.1.86]

$ B = Z t M t W −1 ∂MZ

[11.1.87]

$ In the last equations, C and ∂Care the matrices representing the electrostatic potential and the electric field generated from the electron distribution in the solute molecule, respec$ are the matrices representing the electrostatic potential and tively. The matrices M and ∂M the electric field generated from the nuclear charges in the solute molecule, respectively. The diagonal elements of the matrix W are defined as the following function of the dielectric constant of the solvent W ii =

ε+1 ε −1

[11.1.88]

The subscripts 0 and ∞ in the J1, J2, and T matrices in equation [11.1.82] correspond to the static and optical dielectric constant of the solvent. Equation [11.1.82] can be considered


665

as an analog of equation [11.1.66] for the case of arbitrary cavity shape. The PCM-CI method has been applied for the calculation of solvatochromic shifts in the spectrum 4-[(4’-hydroxyphenyl)azo]-N-methylpyridine in a variety of solvents.48 An integral equation formalism (IEF) has been developed as particularly suitable for the description of solvent effects on spectral transition energies within the PCM model.49 The respective theoretical equations have been applied for the calculation of solvatochromic shifts of several carbonyl-group containing molecules at the self-consistent field (SCF), configuration interaction (CI) and multiconfiguration self-consistent (MC SCF) field level of theory. The calculated spectral shifts accompanying the transfer of a solvatochromic compound from the gas phase to water were comparable with the experimental data. In Table 11.1.4, the results of calculations are presented for three carbonyl compounds, formaldehyde, acetaldehyde and acetone. Table 11.1.4. The calculated and experimental solvatochromic shifts (from the gas phase to water) in the spectra of some carbonyl compounds (cm-1)49 Compound

∆SCF

CI(SDT)

CAS SCF

Exp.

Formaldehyde

1889

839

944

1700-1900a

Acetaldehyde

1854

979

1049

1700-1900a

Acetone

2273

1574

1259

1539-1889

a

an estimate from other compounds

The advantage of the PCM method is in that it is applicable to the solute cavity of practically any shape in the solution. However, it is not clear how precisely should the molecular cavity be defined bearing in mind the classical (quasi-macroscopic) representation of the solvent. It is difficult to perceive that the solvent, e.g., the water molecules, can produce the electrical polarization corresponding to the statistically average distribution in the macroscopic liquid at infinitely small regions on the cavity surface. However, it is conceivable that larger chemical groups in the molecules may possess their own reaction field created by their charge distribution and the reaction fields of other groups in the solute molecule. A multi-cavity self-consistent reaction field (MCa SCRF) has been proposed50 for the description of rotationally flexible molecules in condensed dielectric media. It proceeds from the observation that the interaction of the charge and higher electrical moments of a charge distribution in a spherical cavity with the corresponding reaction fields localized in the center of the cavity does not depend on the position of charge or (point) multipole centers in this cavity. Therefore, it is possible to divide a rotationally flexible solute molecule or a hydrogen-bonded molecular complex between two or more spherical cavities that embed the rotationally separated fragments of the solute or solute and solvent molecules, respectively. Assuming the classical Born-Kirkwood-Onsager charge density expansion (Eq. 11.1.27) for each of these fragments, the total energy of the solute in a dielectric medium can be expressed as a sum of terms that correspond to the energies arising from the interaction of the partial charge and the electric moments of a given molecular fragment with the reaction field of its own and the reaction fields of other fragments, as well as from the interaction between the reaction fields of different fragments. The Hartree-Fock-type equations derived from the variational functional for the total energy E can then be solved iteratively using the

666

Mati Karelson

SCRF procedure. The PCM approach has been further refined to account for the curvature of surface elements.4 Also, this approach has been applied within different quantum-chemical frameworks.47 An alternative method for the description of solute-continuum electrostatic interaction has been developed as based on the notion that the electrostatic equations referring to the boundary surface between the solute and dielectric medium can be substantially simplified if to assume that the solvent is a homogeneous ideally conducting medium. Within this method (called the COSMO method), the electrostatic screening energy of a solute is given by the following equation (in matrix form)51 1 ∆E = − QBA −1BQ 2

[11.1.89]

with the following matrix elements b ik ≈ t k − ri

−1

[11.1.90]

for the point charges and b ik ≈

∫

µ, ν∈ i

χ µ (r)χ ν (r) 3 d r tk −r

[11.1.91]

for the continuous charge distribution, and a kl ≈ t k − t l

−1

, k ≠ l, a kk ≈ 3.8S k

−1/ 2

[11.1.92]

In the last equations, tk denotes the position vectors of the centers of small surface elements k on the arbitrary cavity surface; ri are the position vectors of the point charges in the solute molecule; r is the vector for electronic charge position described on the atomic basis {χ µ ( r )} and Sk are the areas of the surface elements. In equation [11.1.82], Q is the matrix of source charges in the solute. The COSMO model has been extended to account for the solvents with any dielectric constant.52-54 Within the respective GCOSMO method,53 the surface charges σ(r) on the boundary between the solute and continuum solvent are first determined for the medium with the infinite dielectric constant under the assumption that the electrostatic potential on the surface S is zero. For a dielectric medium specified by the dielectric constant ε, the actual surface charges are then calculated by scaling the screening conductor surface charge σ(r) by a factor of f(ε) = (ε - 1)/ε. This scaling preserves the validity of the Gauss theorem for the total surface charge. The boundary element method is applied for the calculation of the surface charge density, with the boundary divided into small areas and the surface charge approximated as the point charge in the center of this area. The charges are calculated either using the charge distribution in the molecule or by minimizing variationally of the total electrostatic solvation energy. The total free energy of the system of solute and surface charges is then calculated within the Hartree-Fock theory as


(

)

(

667

)

1   1 s s 0 0 + + −1 Etot = Pµν Hµν + Hµν + Gµν + Gµν  − 2 f ( ε) Z B A BZ + Enn + Enes [11.1.93] 2 

where Enn is solute nuclear-nuclear repulsion and Enes is the solvation energy related to the dispersion and repulsion between the solute and solvent, and cavity formation; H0µν and G0µν are the one-electron and two-electron parts of the Fock matrix for the isolated solute, respectively, and Pµν are the density matrix elements. The solvent perturbations to the corresponding operators have been expressed as Hsµν = −f (ε)Z+B+ A −1BL µν

[11.1.94]

  G sµν = −f (ε)  ∑ PλσL+λσ A −1L µν  λ, σ 

[11.1.95]

and

where A and B are the N x N square matrices (N - number of atomic nuclei in the solute molecule) with the elements defined by equations [11.1.90] and [11.1.92]. The matrices Lµν consist of the one-electron integrals [11.1.91]. The first and second derivatives needed for the calculation of the molecular potential surfaces and the respective solvent effects on vibrational spectra have been also supplied within the framework of GCOSMO approach.53 A semi-quantitative agreement between the computational and experimental results has been obtained for the vibrational frequencies of acetone in water. Several approaches have been developed to account for the electron correlation effects on the solvation energy of both the ground and the excited states of the molecule in the solution. A multiconfigurational self-consistent reaction field (MC SCRF) theory has been proposed as based on the classical Onsager’s reaction field model.55 Notably, the higher order electrical moments of the solute molecule and the respective reaction field in the solvent were taken into account within this method. Thus, the dielectric solvation energy of a solute in a given state embedded into a linear isotropic medium has been calculated as the product of the expectation values for the reaction field ¢R ml ¦ and the respective multipole charge moments ¢M ml ¦ of the solute55-58 E el = −

+ 1 R ml M ml ∑ 2 l, m

[11.1.96]

where M ml = ∑ Z k S lm (Rk ) − S lm

[11.1.97]

R ml = f M ml

[11.1.98]

k

and

668

Mati Karelson

fl =

(l + 1)(ε − 1) 1 1 + ε(l + 1) a − ( 2 l+1)

 4π  S lm (r) =   2l + 1

1/ 2

r lY lm (θ, ϕ)

[11.1.99]

[11.1.100]

where Ylm(θ, ϕ) are the Legendre’ polynomials. Interestingly, it had been suggested that the dispersion interaction energy between the solute and the solvent could be accounted for, in principle, in the framework of this approach as related to the full distribution of dielectric relaxation frequencies of the solvent. Thus, the formula for the MC SCRF solvation energy has been expressed as follows E el = −

1 2 f l (Tlm ) ∑ 2 l, m

[11.1.101]

with the terms Tlm obtained from the expectation values of the nuclear and electronic solvent operators Tlm = Tlmn − Tlme

[11.1.102]

Tlmn = ∑ Z a R lm (R a )

[11.1.103]

Tlme = R lm (r)

[11.1.104]

a

where Rlm are the special solvent effect integrals.55 The solvent contributions have been developed also for the multiconfigurational energy gradient and Hessian, necessary for the solution of the MC SCRF equations. Notably, the results of the model calculations on water molecule implied that the higher multipole terms might play a significant role in the total electrostatic solvation energy of the molecule. Thus, the quadrupole term consisted approximately 20% of the dipolar term whereas the 4-th order term was even more significant (~ 30% of the dipolar term). The MC SCRF method has been applied for the calculation of the solvent effects on the spectral transitions of water and formaldehyde in different media.59 The MC SCRF methodology has been further refined in the framework of the response theory approach.60 This development describes the response of the solute or the solute-solvent complex to a time-independent or time-dependent high-frequency perturbation such as the spectral excitation. In order to take into account the electron correlation effects, another combination of the self-consistent reaction field theory with the configuration interaction formalism has been introduced as follows.61 Within this approach, the usual CI wavefunction has been constructed as follows N

ΨCI = ∑ Ca Da a =1

[11.1.105]


669

where Ca are the CI expansion coefficients and Da are the basis functions (Slater determinants or their linear combinations). In the case of orthogonal basis functions, the normalization condition of the function ΨCI is given as N

∑C a =1

2 a

=1

[11.1.106]

and the coefficients Ca are determined from the following equation

(H$

0

)

+ V$rf Ψ = EΨ

[11.1.107]

$ is $ 0 is the Hamiltonian for the molecule, unperturbed by the reaction field and V where H rf the reaction field perturbation. The latter can be presented, for example, in the framework of the boundary element method as follows (cf. Eq. [11.1.75]) σ(r ′) V$rf = ∫ d 2 r r − r′

[11.1.108]

where σ( r ′ ) is the charge density on the surface of the cavity. For the spectroscopic applications, it would be again instructive to separate the noninertial and inertial components of the electrostatic polarization of the dielectric medium. The first of them corresponds to the electrostatic polarization of the electron charge distribution in the solvent that is supposedly instantaneous as compared to any electronic or conformational transition of the solute. The second component arises from the orientational polarization of the solvent molecules in the electrostatic field of the solute. The noninertial polarization can be described by the optical dielectric permittivity of the solvent that corresponds to the infinite frequency of external electromagnetic field (ε ∞ ≈ n D2 ) whereas the inertial polarization represents the slow, orientational part of the total dielectric constant of the solvent, ε. In order to separate the noninertial polarization, it is helpful to determine the solute charge density as the sum of the respective nuclear and electronic parts ρ = ρ n + ρ e = ∑ δ(r − rA )Z A + ∑ Ca Cb (ab | r, r ′) = ∑ Ca Cb ρ ab [11.1.109] A

a, b

a, b

where δ(r - rA) is the Dirac’s delta function and (ab|r,r') are the elements of the single-determinant matrices of transitions between the configurations. Notably, the values of ρ ab = δ ab ρ n − ρ(ab | r, r ′)

[11.1.110]

do not depend on the coefficients Ca and Cb. The noninertial component of the polarization field, Φ ∞ ( r ), is always in equilibrium and thus it can be represented as follows Φ ∞ (r) = ∑ Ca Cb Φ (ab∞ ) (r)

[11.1.111]

a, b

where Φ (ab∞ ) ( r ) is the solution of equation [11.1.107] for the basic charge distribution ρab. Since the latter does not depend on the coefficients Ca and Cb, the values of Φ (ab∞ ) ( r ) can be

670

Mati Karelson

determined before the calculation of the wavefunction ΨCI [11.1.105]. The coefficients Ca can be calculated, as usual, from the matrix equation H|C¦ = E|C¦

[11.1.112]

where |C¦ denotes the column-vector of the coefficients and the elements of the matrix H are given as follows61 $ 0 D + C C d 3 rρ Φ ( ∞ ) + d 3 rρ Φ Hab = Da H ∑ a b∫ 0 b ab ab ab ∫

[11.1.113]

a, b

where Φ0 is inertial (nuclear) part of the polarization field. For a given Φ0, the set of equations [11.1.98) can be solved iteratively. Implicitly, the last equations describe both the electron subsystems of the solute and solvent, the latter being taken into account as the field of noninertial (electron) polarization of the solvent, Φ ( ∞ ) . The electron correlation effects on the solvation energy of a solute have been also accounted for within the framework of the perturbation theory.62,63 By starting from the Hamiltonian of the solute molecule as follows $ =H $ 0 − M mf m ψ M m′ ψ H l l l′

[11.1.114]

where ψ denotes the exact (correlated) wavefunction, the Hartree-Fock operator may be written as F = F 0 − M lm f l m ψ 0 M lm′ ′ ψ 0

[11.1.115]

where ψ 0 denotes the electronic wavefunction at the Hartree-Fock level. The Hamiltonian may be then written as a perturbed expression of the Hartree-Fock operator

(

$ =F + H $ 0 − F0 H

)+M f ( ψ m m l l

0

M lm′ ′ ψ 0 − ψ M lm′ ′ ψ

)

[11.1.116]

The perturbation has two contributions, the standard Møller-Plesset perturbation and the non-linear perturbation due to the solute-solvent interaction. If C(i)j denotes the coefficient of the eigenstate |J¦ in the corrections of ψ to the i-th order, then the perturbation operators H(i) of the i-th order are given by the following formulae $ (1) = H $ 0 − F0 H

[11.1.117]

$ ( 2 ) = −2 C (1) M m f m 0M m ′ I H ∑ I l l l′

[11.1.118]

I≠0

$ ( 3 ) = −2 C ( 2 ) M m f m 0M m ′ J − H ∑ J l l ∑ ∑ CJ(1)CP(1)M lmf l m I M lm′ ′ P l′ J ≠0

J ≠0 P ≠0

[11.1.119]


671

The calculation is performed, as usual, by comparing the coefficients of the Schrödinger equation to successive orders. The first order energy is the same as given by the usual Møller-Plesset treatment, $ 0 |0 > and the first order electrostatic contribution to the free energy of solvation is < 0| H identical to the result obtained at the Hartree-Fock level of theory. The second order correction to the free energy is given as ∆G s(2 ) = 2∑CS(2 ) 0M lm′ ′ S fl m 0M lm′ ′ 0 + S≠0

∑ ∑C

D≠0 D ′≠0

CD(1)′ D M lm′ ′ D ′ fl m 0M lm′ ′ 0

(1) D

[11.1.120]

where |S¦ stands for the singly excited states, and |D¦ and |D'¦ for a pair of doubly excited states different by just one orbital. Without excessive difficulty, it is possible to derive the correction terms to the electrostatic free energy of solvation of higher orders. A many-body perturbation theory (MBPT) approach has been combined with the polarizable continuum model (PCM) of the electrostatic solvation.64-66 The first approximation called by authors the perturbation theory at energy level (PTE) consists of the solution of the PCM problem at the Hartree-Fock level to find the solvent reaction potential and the wavefunction for the calculation of the MBPT correction to the energy. In the second approximation, called the perturbation theory at the density matrix level only (PTD), the calculation of the reaction potential and electrostatic free energy is based on the MBPT corrected wavefunction for the isolated molecule. At the next approximation (perturbation theory at the energy and density matrix level, PTED), both the energy and the wave function are solvent reaction field and MBPT corrected. The self-consistent reaction field model has been also applied within the complete active space self-consistent field (CAS SCF) theory12,67 and the complete active space second-order perturbation theory.12,67,68 Several groups69-73 have also proposed the quantum mechanical density functional theory (DFT) based methods for the calculation of the electrostatic solvation energy in dielectric media. However, the application of this theory for excited states is not straightforward.74,75 11.1.4 THEORETICAL TREATMENT OF SOLVENT DISPERSION EFFECTS ON ELECTRONIC-VIBRATIONAL SPECTRA OF MOLECULES The dispersion interaction between two atomic or molecular systems can be theoretically presented at different levels of theory.76-78 The modelling of the dispersion interactions in condensed media is more complicated and proceeds either from the discrete molecular description of the liquid or from the continuum model. According to a contemporary classification,4 the theoretical approaches to the dispersion effect in solutions can be divided into following classes: • pair-potential approaches • reaction field based approaches • cavity surface-dispersion energy relationship approaches The pair-potential approach is based on the discrete representation of the pairs of solvent and solute molecules or some fragments of them. The respective dispersion potentials are expressed as truncated asymptotic expansions in powers of 1/r, the reciprocal of the distance between the interacting entities4 Ums (disp) =

∑d

k = 6, 8,10

k −k ms ms

r

[11.1.121]

672

Mati Karelson

where the indexes m and s denote the structural entities (atoms, bonds, chemical groups) belonging to the solute and solvent molecules, respectively. The powers in expansion [11.1.121] are based on the formal theory of two-body interactions. In the practical calculations, only the first term of the expansion (k = 6) is frequently applied. The expansion coefficient d (6) ms can be calculated using the London formula I I 3 (6 ) d ms = − α m αs m s 2 I m + Is

[11.1.122]

where αm and αs are the isotropic polarizabilities for interacting systems and Im and Is are the mean excitation energies of these systems. This approximate formula is, in principle, valid only for interacting atoms. In the case of molecular systems, the atomic or group polarizabilities and local excitation energies are, as a rule, not isotropic and require the use of the respective tensor quantities. The absence of information about of accurate solute-solvent atom-atom distribution functions in dense media complicates further the accurate treatment of the dispersion interaction. These distribution functions can be calculated either using the molecular dynamics or Monte Carlo computer simulations or from the experimental scattering data on the respective systems. However, almost all these methods give only the averaged distribution functions and lack, therefore, the information about the local anisotropy of the atom-atom distributions. Similarly to the treatment of electrostatic effects, the dispersion potential can be limited to the dipole-dipole term and the mean excitation energies are approximated by the respective ionization potentials for the solute and solvent molecules. Thus, when a small cluster of solvent molecules surrounds the solute molecule, the first approximation of the dispersion energy can be presented by the following formula:79 UMS (disp) = −

x I M IS 4 I M + IS

BM

BS

∑ ∑ {r u =1 v =1

}

Tr [Tuv A u Tuv Av ]

−6 uv

[11.1.123]

where BM and BS are the number of bonds in the solute and in the solvent molecules, respectively, Im and Is are the corresponding mean excitation energies, Tuv is the tensor Tuv = 3

ruv r ⊗ uv − 1 ruv ruv

[11.1.124]

where ruv and ruv are the distance and the radius-vector between the bonds u and v, respectively, and Au is the polarizability tensor for bond u. The factor x in equation [11.1.123] is introduced to achieve the agreement between the molecule-molecule pair dispersion potential and a simpler expression derived on the basis of assumption that the dispersion energy between two molecules may be reduced to the sum of independent atom-atom contributions80 ( 6 ) −6 UMS (disp) = ∑ ∑ d ms rms m

s

[11.1.125]


673

A scheme has been developed that reduces the spatial representation of the dispersion interaction into a surface representation of this interaction.4 According to this approach, the average dispersion-repulsion energy of a solute-solvent system has been written as follows: E disp − rep = ∫ L ∫ U(Ω) g (Ω)dΩ

[11.1.126]

where Ω stands for the set of all coordinates of the molecules involved, g(Ω) is the solute-solvent pair distribution function and U(Ω) is expressed as a sum of two-body dispersion-repulsion potentials. In the case of the fixed geometry of the solute molecule E disp − rep = nS ∑ NS s∈S

∑ ∑d ∫ r (k ) ms

m∈ M k

(k ) ms

3 g ms (rms )drms

[11.1.127]

The integrals in the last formula can be limited only to a certain minimum distance defined, for instance, by the van-der-Waals envelopes of interacting molecules. By introducing the auxiliary vector functions A (k) ms ( r ms ) such that r (k ) (k ) − k [11.1.128] rms g ms (rms ) ∇Ams (rms ) = d ms the average dispersion-repulsion energy between the solute and solvent molecules in solution may be written as follows E disp − rep = nS ∑ NS s∈S

∑∑ ∫A

m∈ M k

(k ) ms

n σdσ

[11.1.129]

Σs

where nσ is the outer normal to the surface Σs at the position σ. The integral in the last equation may be calculated numerically using an appropriate partitioning (tessellation) of the surface. A quantum-mechanical method of calculation of the dispersion energy has been developed on the basis of the above-cited semiclassical Abe’s theory.81 According to this method, the dispersion energy, Edisp , for a solute molecule in a spherical cavity is given as follows E disp

2 1 =− 3 aS3 a M3

∑ ∑E J ≠ I K ≠O

(µ ) (µ ) M IJ

S K

2

S KO

2

− E OS + E JM − E IM

[11.1.130]

where the superscript S refers to the solvent molecule and the superscript M to the solute S are the transition dipoles between the respective states of the molecule. Thus, µ IJM and µ KO solute (I and J) and the solvent (K and O) molecules. In equation [11.1.130], EKS , E SO and M EM J , EI denote the energies of the K-th and O-th state of the solvent and of the J-th and I-th state of the solute molecule, respectively. The cavity radii for the solvent and solute molecules are denoted as aS and aM, respectively.

674

Mati Karelson

Table 11.1.5. The INDO/CI calculated solvatochromic shifts ∆ν (from the gas phase to cyclohexane) of some aromatic compounds and the respective experimental data in low polarity solvents (cm-1)81 Compound

∆ν (calc)

∆ν (exp)

1

B2u

-316

-209a

Lb(x) Lb(y)

-332 -879

-300b; -275a -950b; -902a

Lb(1Bu) La(1Bu) 1 Bb(1Bu)

-243 -733 -1666

-252a -1030a -1620a

+162 -288 -446 -1475

+164c -333c -285c -1650c

Transition

Benzene 1

Naphthalene

1 1

Chrysene

1

1

Lb(y) La(x) 1 Kb(y) 1 Bb(x) 1

Azulene

a

in n-pentane, bin cyclohexane, cin 2-chloropropane

The equation [11.1.130] has been used within the semiempirical quantum-chemical INDO/CI formalism to calculate the solvent shifts of some aromatic compounds in cyclohexane.81 The results compare favorably with the experimental data for some nonpolar solvents (cf. Table 11.1.5). 11.1.5 SUPERMOLECULE APPROACH TO THE INTERMOLECULAR INTERACTIONS IN CONDENSED MEDIA The supermolecule approach to the calculation of solute-solvent interaction energies is based on the discrete molecular representation of the solvent. The supermolecule can be treated quantum-mechanically as a complex consisting of the central solute molecule and the surrounding closest solvent molecules. This supermolecule complex can be treated individually or as submerged into the dielectric continuum.82 In the last case, some continuum theory (SCRF, PCM) is applied to the supermolecule complex consisting of the solute molecule and the solvent molecules in its first coordination sphere.83-86 Therefore, the short-range solute-solvent electron correlation, dispersion and exchange-repulsion interactions are taken into account explicitly at the quantum level of theory as the electrons and nuclei both from the solute and solvent are included explicitly in the respective Schrödinger equation. The long-range electrostatic polarization of the solvent outside the first coordination sphere is, however, treated according to the dielectric continuum theory. Thus, the energy of solvation of a solute molecule can be expressed as follows: $ (S ) Ψ $ (0 ) Ψ $ (0 ) Ψ E sol = ΨSM H − ΨSM H − n ΨSM H SM SM SM SM M S

[11.1.131]

$ (0) and H $ (0) are $ (S) is the Hamiltonian for the supermolecule in the solution, and H where H SM M S the Hamiltonians for the isolated solute and the solvent molecules, respectively. In the last equation, n denotes the number of the solvent molecules applied in the supermolecule, ΨSM is the total wavefunction of the supermolecule immersed into dielectric medium, and ΨM and ΨS are the wavefunctions for isolated solute and solvent molecules, respectively.


675

For instance, the n → π * electronic transition (11A1 →11A2) of formaldehyde solvated by varying number of water molecules has been investigated using multi-reference CI calculations.87 This simple supermolecule approach has given already satisfactory results as compared to experimental shifts in liquid water. However, it has been shown that in general, both the short-range quantum mechanical effects and the long-range solvent polarization play important role in determining the spectral shifts in liquid media. Thus, the INDO/S SCRF CIS theory alone has explained the solvatochromic shifts of azoles in different solvents, except those observed in water (Table 11.1.6).85 In both the water and acetonitrile, the compounds are predicted to have practically the same shift that is not the case in experiment. The explicit bonding of two water molecules to the nitrogen lone pairs leads in the cases of pyrimidine and pyridazine to the calculated large red shift instead of the experimentally observed solvatochromic blue shift. However, by treating the complex of an azole and two water molecules quantum-mechanically in the surrounding reaction field leads to quantitatively correct blue shifts. Table 11.1.6. The INDO/S SCRF/CI calculated and experimental spectral transition energies in different solvents for azoles (cm-1)85 νcalc, cm-1

νexp, cm-1

Pyrimidine

gas phase isooctane diethyl ether acetonitrile water 2H2O water + 2H2O

32966 33559 34127 34697 34743 30982 36572

34200 34400 34800 36900 36900 36900

Pyridazine


28329 29460 30382 31296 31368 26490 33927

29740 30150 31080 33570 33570 33570

Pyrazine


30387 30387 30387 30387 30387 32900 33301

31610 31610 31740 33160 33160 33160

Molecule

Solvent

As a general remark, in the calculations of the intermolecular interactions using the supermolecule approach, the “size-extensivity”88 of the methods applied is of crucial importance. Furthermore, the interaction energies calculated in the supermolecule approach usually suffer from what is called the basis set superposition error (BSSE),89 a spurious energy improvement resulting from the use of truncated basis sets. This error seems to be unavoidable in most practical calculations except for very small systems.90

676

Mati Karelson

The intermolecular interactions that correspond to the fixed geometry of the solute-solvent complex can be also studied by a perturbation approach.91,92 It has been suggested that the perturbation theory has some advantages over the supermolecule approach and may therefore be considered conceptually more appropriate for the calculation of intermolecular interaction energies. In this case, the interaction energy is calculated directly and it may be separated into components of well-defined physical meaning. Within the direct reaction field (DRF) method,93-96 the classical part of the solute-solvent system (solvent) is treated as a distribution of the polarizable point dipoles, interacting with each other. The DRF Hamiltonian of the solute-solvent system is thus given by the following formula: $ $0 1 H ∑ ∑ Fip+ α pq Fjq DRF = H − 2 i, j p, q

[11.1.132]

where indices i an j correspond to the solute particles (electrons and nuclei) and p and q run over the external polarizable points. Fip is the field of the particle i at the position p, and αpq gives the induced dipole at point q by a field applied at point p. The respective Schrödinger equation can be solved directly, without the iterative adjustment of the solvent charge distribution and the respective reaction field potential. The DRF method proceeds from the direct reaction field obtained as the linear solute-solvent interaction operator, proportional to the square of the electric field operator while the GSCRF approach uses the average reaction field model. It has been suggested that the additional energy contributions can be interpreted as due to the dispersion interaction between the solute and solvent molecules.97 More recently, the DRF approach has been combined with the continuum approach by dividing the space around the solute into a closer surrounding treated by direct reaction field method and to more distant space represented by the macroscopic dielectric properties of the solvent.98,99 A good quantitative agreement has been obtained between the experimental and DRF calculated solvatochromic shifts of the n → π * transition of acetone in different solvents.100,101 Nevertheless, even at the highest theoretical level accessible for practical calculations, the static approach is strictly valid only for the description of the molecular clusters of fixed geometry. However, in the cases of strong and weak intermolecular interactions, the energy of interaction in the molecular cluster in the gas phase or on the inert-gas matrix is substantially different from the total solvation energy in the condensed phase.102 A direct solution of a time-dependent Schrödinger equation for the condensed low-order bulk matter, needed to overcome this problem, is premature. Therefore, the molecular dynamics method (MD)103-105 based on the computer modelling of a system of molecules which interact by the known model potential to each other and undergoes the rotational and translational movement in the field caused by this interaction according to the classical (Newtonian) mechanics is widely applied for this purpose. By applying various boundary conditions and performing the calculation of the potential energies and forces for hundreds of thousands configurations obtained by step-by-step time evolution the time - averages such as internal energy (or enthalpy) can be obtained. An alternative is the stochastic Monte Carlo method that is based on the ergodic theorem and provides the ensemble averages calculated from randomly generated and weighted configurations. These methods suffer from several shortcomings, of which the problem of the applicability of the ergodic theorem (i.e., the identity


677

of the time-averaged and ensemble-averaged thermodynamic and dynamic observables), the path sampling and difficulties to obtain precise intermolecular interaction potentials are most serious. Also, the real dense systems, i.e., liquids and solutions are intrinsically quantified systems and therefore a quantum molecular dynamics should be developed which accounts for the quantum effects in the microscopic system from the first principles. The combined quantum-mechanical/molecular dynamics (QM/MD) or quantum-mechanical/molecular mechanics (QM/MM) approaches have been used for the calculation of solvatochromic shifts in different media.106-108 Numerous computational schemes have been developed to calculate the total molecular solvation energy or free energy using the combination of different theoretical solute-solvent interaction models. From these, one of the most popular is the SMx methodology.109,110 This methodology proceeds from the division of the total molecular solvation energy into the solute-solvent electrostatic and inductive polarization terms, standard-state free energy of cavity creation in the solvent plus the solute-solvent dispersion interaction, and an empirical part of the nuclear motion free energy. The solvent polarization term is presented using the generalized Born formula: 1  1 N N G p = −  1 − ∑ ∑ q k q k ′ γ kk ′ 2  ε  k =1 k ′=1

[11.1.133]

where the double summation is performed over all atomic partial charges qk in the solute molecule, ε is the relative dielectric permittivity of the solvent and γkk’ - the Coulomb’ integral between two centers k and k’, parameterized for the interactions with the solvent. The cavity creation plus dispersion term is calculated as N

0 = ∑ σ k Ak GCD

[11.1.134]

k =1

where N is the number of atoms in the solute, Ak is the solvent accessible surface area of a given atom and σk is the parameter for this atom that is called the accessible surface tension. The latter is obtained from the fit with the experimental data and is, thus, essentially an empirical parameter for a given type of atom. Thus, in essence the SMx methodology represents a semiempirical approach to the calculation of solvent effects. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14

P. Suppan, N. Ghoneim, Solvatochromism, The Royal Society of Chemistry, London, 1997. N.G. Bakhshiev (Ed.), Solvatochromism - Problems and Methods (in Russian), Leningrad Univ. Publ. House, Leningrad, 1989. M. Karelson, Adv. Quant. Chem., 28, 142 (1997). J. Tomasi, M. Persico, Chem. Rev., 94, 2027 (1994). T. Simonson, A.T. Brünger, J. Phys. Chem., 98, 4683 (1994). O. Sinanoglu, Chem. Phys. Lett., 1, 283 (1967). V. Gogonea, E. Osawa, J. Mol. Struct. (THEOCHEM), 311, 305 (1994). H. Reiss, Adv. Chem. Phys., 9, 1 (1966). R.A. Pierotti, Chem. Rev., 76, 717 (1976). R.M. Gibbons, Mol. Phys., 17, 81 (1969). M. Irisa, K. Nagayama, F. Hirata, Chem. Phys. Lett., 207, 430 (1993). A. Bernhardsson, R. Lindh, G. Karlström, B.O. Roos, Chem. Phys. Lett., 251, 141 (1996). W. Jaskólski, Phys. Rep., 271, 1 (1996). S. Basu, Adv. Quant. Chem., 1, 145 (1964).

678

15 16 17 18 19 20 21 22 23 24 25 26

27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

Mati Karelson A.T. Amos, B.L. Burrows, Adv,. Quant. Chem., 7, 289 (1973). R.J. Abraham, B.D. Hudson, M.W. Kermode, J.R. Mines, J. Chem. Soc. Farad. Trans., 84, 1911 (1988). J.G. Kirkwood, J. Chem. Phys., 7, 911 (1939). L. Onsager, J. Am. Chem. Soc., 58, 1486 (1936). C.J.F. Böttcher and P. Bordewijk, Theory of Electric Polarization, 2nd ed., vol. 2, Elsevier Co, Amsterdam, 1978. D. Rinaldi, M.F. Ruiz-Lopez, J.-L. Rivail, J. Chem. Phys., 78, 834 (1983). D. Rinaldi, J.-L. Rivail, N. Rguini, J. Comput. Chem., 13, 675 (1992). J.-L. Rivail, B. Terryn, D. Rinaldi, M.F. Ruiz-Lopez, J. Mol. Struct. (THEOCHEM), 166, 319 (1988). E.G. McRae, J. Phys. Chem., 61, 562 (1957). T. Abe, Bull Chem. Soc. Japan, 38, 1314 (1965). O. Tapia, O. Goscinski, Mol. Phys., 29, 1653 (1975). M. Karelson, G.H.F. Diercksen, “Models for Simulating Molecular Properties in Condensed Systems, in: Problem Solving in Computational Molecular Science. Molecules in Different Environments, S. Wilson and G.H.F. Diercksen (Eds.), Kluwer Academic Publishers, Dordrecht, 1997, pp. 215- 248. M. Karelson, M.C. Zerner, J. Phys. Chem., 92, 6949 (1992). R. Marcus, J. Chem. Phys., 24, 979 (1956). J. Jortner, Mol. Phys., 5, 257 (1962). J. Ridley, M.C. Zerner, Theor. Chim. Acta, 42, 223 (1976). A.D. Bacon, M.C. Zerner, Theor. Chim. Acta, 53, 21 (1979). M. Karelson, K. Pihlaja, T. Tamm, A.Uri, and M.C. Zerner, J. Photochem. and Photobiol. A Chemistry, 85, 119 (1995). A. Broo, Chem. Phys., 183, 85 (1994). A. Broo, M.C. Zerner, Chem. Phys. Lett., 227, 551 (1994). J.-L. Rivail, D. Rinaldi, V. Dillet, Mol. Phys., 89, 1521 (1996). M.M. Karelson, Org. React., 17, 357 (1980). M.W. Wong, K.B. Wiberg, M.J. Frisch, J. Am. Chem. Soc., 113, 4776 (1991). O. Tapia, F. Colonna, J. Ángyán, J. Chim. Phys., 87, 875 (1990). G. Jansen, F. Colonna, J. Ángyán, Int. J. Quant. Chem., 58, 251 (1996). S. Miertuš, E. Scrocco, J. Tomasi, Chem. Phys., 55, 117 (1981). H. Hoshi, M. Sakurai, Y. Inoue, R. Chûjô, J. Chem. Phys., 87, 1107 (1987). M.L. Drummond, J. Chem. Phys., 88, 5014 (1988). B. Wang, G.P. Ford, J. Chem. Phys., 97, 4162 (1992). T. Fox, N, Rösch, R.J. Zauhar, J. Comput. Chem., 14, 253 (1993). F.J. Luque, M. Orozco, P.K. Bhadane, S.R. Gadre, J. Chem. Phys., 100, 6718 (1994). C. Gonzalez, A. Restrepo-Cossio, M. Márquez, K.B. Wiberg, M. De Rosa, J. Phys. Chem. A, 102, 2732 (1998). M.A. Aguilar, J. Olivares del Valle, J. Tomasi, J. Chem. Phys., 98, 7375 (1993). H. Houjou, M. Sakurai, Y. Inoue, J. Chem. Phys., 107, 5652 (1997). B. Mennucci, R. Cammi, J. Tomasi, J. Chem. Phys., 109, 2798 (1998). M. Karelson, T. Tamm, M.C. Zerner, J. Phys Chem., 97, 11901 (1993). A. Klamt, G. Schüürmann, J. Chem. Soc. Perkin Trans. 2, 799 (1993). A. Klamt, J. Phys. Chem., 99, 2224 (1995). E.V. Stefanovich, T.N. Truong, J. Chem. Phys., 105, 2961 (1996). A. Klamt, J. Phys. Chem., 100, 3349 (1996). K.V. Mikkelsen, H. Ågren, H.J. Aa Jensen, T. Helgaker, J. Chem. Phys., 89, 3086 (1988). J.L. Rivail, D. Rinaldi, Chem. Phys., 18, 233 (1976). M.D. Newton, J. Phys. Chem., 79, 2795 (1975) K.V. Mikkelsen, E. Dalgaard, P. Swanstrom, J. Phys. Chem., 91, 3081 (1987). K.V. Mikkelsen, A. Cesar, H. Ågren, H.J. Aa. Jensen, J. Chem. Phys., 103, 9010 (1995). K.V. Mikkelsen, P. Jørgensen, H.J. Aa. Jensen, J. Chem. Phys., 100, 6597 (1994). M.V. Basilevsky, G.E. Chudinov, Chem. Phys., 157, 245 (1991). J.-L. Rivail, C. R. Acad. Sci. Paris, 311 II, 307 (1990). J.-L . Rivail, D. Rinaldi, M.F. Ruiz-Lopez, in: Theoretical and Computational Models for Organic Chemistry; S.J. Formosinho, I.G. Csizmadia, and L.G. Arnaut, (Eds.), Kluwer, Dordrecht. 1991, pp. 77-92. F.J. Olivares del Valle, J. Tomasi, Chem. Phys., 150, 139 (1991). F.J. Olivares del Valle, M.A. Aguilar, J. Comput. Chem., 13, 115 (1992). F.J. Olivares del Valle, M.A. Aguilar, J. Mol. Struct. (THEOCHEM), 280, 25 (1993).


67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105

106 107 108 109 110

679

N.A. Besley, J.D. Hirst, J. Phys. Chem. A, 102, 10791 (1998). L. Serrano-Andrés, M.P. Fülscher, G. Karlström, Int. J. Quant. Chem., 65, 167 (1997). R.R. Contreras, A.J. Aizman, Int. J. Quant. Chem. Quantum Chem. Symp., 25, 281 (1991). R.J.Hall, M.M. Davidson, N.A. Burton, I.H. Hillier, J. Phys. Chem., 99, 921 (1995). R.R. Contreras, P. Pérez, A.J. Aizman, Int. J. Quant. Chem., 56, 433 (1995). G.J. Tawa, R.L. Martin, L.R. Pratt, T.V. Russo, J. Phys. Chem., 100, 1515 (1996). T.N. Truong, E.V. Stefanovich, Chem. Phys. Lett., 240, 253 (1995). E.K.U. Gross, W. Kohn, Phys. Rev. Lett., 55, 2850 (1985). C.A. Ullrich, E.K.U. Gross, Phys. Rev. Lett., 74, 872 (1995). H. Margenau, N.R. Kestner, Theory of Intermolecular Forces, Pergamon Press, Oxford, 1971. J. Mahanty, W.B. Ninham, Dispersion Forces, Academic Press, New York, 1976. P. Hobza, P. Zahradnik, Intermolecular Complexes, Elsevier, Amsterdam, 1988. M.J. Huron, P. Claverie, Chem. Phys. Lett., 4, 429 (1969). A.J. Pertsin, A.I. Kitaigorodsky, The Atom-Atom Pair Potential Method, Springer, Berlin, 1986. N. Rösch, M.C. Zerner, J. Phys. Chem., 98, 5817 (1994). G.H.F. Diercksen, M. Karelson, T. Tamm and M.C. Zerner, Int. J. Quant. Chem., S28, 339 (1994). M.M. Karelson, Org. React., 20, 127 (1983). O. Tapia, J.M. Lluch, R. Cardenas, J. Andres, J. Am. Chem. Soc., 111, 829 (1989). M.M. Karelson, M.C. Zerner, J. Am. Chem. Soc., 112, 9405 (1990). L.C.G. Freitas, R.L. Longo, A.M. Simas, J. Chem. Soc. Faraday Trans., 88, 189 (1992). I. Frank, S. Grimme, M. von Arnim, S.D. Peyerimhoff, Chem. Phys., 199, 145 (1995). E. Clementi, G. Corongiu, (Eds.) Computational Chemistry, METECC-95, STEF, Cagliari, 1995. S.F. Boys, F. Bernardi, F. Mol. Phys., 19, 553 (1970). Z. Latajka, S. Scheiner, J. Comput. Chem., 8, 674 (1987). P. Arrighini, P. Intermolecular Forces and Their Evaluation by Perturbation Theory, Lecture Notes in Chemistry, vol. 25, Springer-Verlag, Berlin, 1981. B.J. Jeziorski, R. Moszynski, A. Ratkiewicz, S. Rybak, K. Szalewicz, H.L. Williams, Chapter 3 in: Methods and Techniques in Computational Chemistry, METECC-94, vol. B. STEF, Cagliari, 1994. B.T. Thole, P.Th. van Duijnen, Chem. Phys., 71, 211 (1982). B.T. Thole, P.Th. van Duijnen, Theor. Chim. Acta, 63, 209 (1983). B.T. Thole, P.Th. van Duijnen, Biophys. Chem., 18, 53 (1983). P.Th. van Duijnen, A.H. Juffer, H. Dijkman, H. J. Mol. Struct. (THEOCHEM), 260, 195 (1992). A.H. deVries, P.Th. van Duijnen, Int. J. Quant. Chem., 57, 1067 (1996). H. Dijkman, P. Th. van Duijnen, Int. J. Quant. Chem. Quantum Biol. Symp., 18, 49 (1991). P.Th. van Duijnen, A.H. de Vries, Int. J. Quant. Chem. Quantum Chem. Symp., 29, 531 (1995). A.H. de Vries, P.Th. van Duijnen, Int. J. Quant. Chem., 57, 1067 (1996). F.C. Grozema, P.Th. van Duijnen, J. Phys. Chem. A, 102, 7984 (1998). F.H Stillinger, Adv. Chem. Phys., 31, 1 (1975). S. Engström, B. Jönsson, R.W. Impey, J. Chem. Phys., 80, 5481 (1984). W.F. van Gunsteren, Current Opinion in Structural Biology, 3, 277 (1993). C. Corongiu, V. Martorana, Chapter 3 in: Methods and Techniques in Computational Chemistry, METECC-94; vol. C. STEF, Cagliari, 1994.; W.R.P. Scott, W.F. van Gunsteren, Chapter 9 in: Methods and Techniques in Computational Chemistry, METECC-95; Ed. by E. Clementi and G. Corongiu, STEF, Cagliari, 1995. J.T. Blair, K. Krogh-Jespersen, R.M. Levy, J. Am. Chem. Soc., 111, 6948 (1989). V. Luzhkov, A. Warshel, J. Am. Chem. Soc., 113, 4491 (1989). J.S. Bader, C.M. Cortis, B.J. Berne, J. Chem. Phys., 106, 2372 (1997). C.J. Cramer, D.G. Truhlar, J. Am. Chem. Soc., 113, 8305 (1991). C.J. Cramer, D.G. Truhlar, J. Am. Chem. Soc., 113, 8552 (1991).

680

Tai-ichi Shibuya

11.2 DIELECTRIC SOLVENT EFFECTS ON THE INTENSITY OF LIGHT ABSORPTION AND THE RADIATIVE RATE CONSTANT Tai-ichi Shibuya Faculty of Textile Science and Technology Shinshu University, Ueda, Japan

11.2.1 THE CHAKO FORMULA OR THE LORENTZ-LORENZ CORRECTION The intensity of light absorption by a molecule is generally altered when the molecule is immersed in a solvent or transferred from one solvent to another. The change may be small if the solvents are inert and non-polar, but often a significant increase or decrease is observed. The first attempt to correlate such effects with the nature of the solvent was made by Chako1 in 1934. Chako’s formula reads as

(

)

n2 + 2 f ′′ = 9n f

2

[11.2.1]

where: f f′ n

oscillator strength of an absorption band of a molecule apparent oscillator strength of the molecule in solution refractive index of the solution at the absorbing frequency

The apparent oscillator strength is proportional to the integrated intensity under the molar absorption curve. To derive the formula, Chako followed the classical dispersion theory with the Lorentz-Lorenz relation (also known as the Clausius-Mosotti relation), assuming that the solute molecule is located at the center of the spherical cavity in the continuous dielectric medium of the solvent. Hence, the factor derived by Chako is also called the Lorentz-Lorenz correction. Similar derivation was also presented by Kortòm.2 The same formula was also derived by Polo and Wilson3 from a viewpoint different from Chako. Chako’s formula always predicts an increase of the absorption intensity with the refractive index. This does not hold, for instance, for the allowed π → π * electronic transitions of cyclohexadiene and cyclopentadiene,4 and monomethyl substituted butadienes.5 11.2.2 THE GENERALIZED LOCAL-FIELD FACTOR FOR THE ELLIPSOIDAL CAVITY A natural generalization of the Chako formula was made by generalizing the spherical cavity to an ellipsoidal cavity. Such a generalization was shown by Shibuya6 in 1983. The generalized formula derived by him reads as

[(

) ]

s n2 −1 + 1 f ′′ = f n

2

[11.2.2]

where: s

shape parameter which takes a value between 0 and 1

This parameter s is more generally known as the depolarization factor, whose values are listed for special cases in general textbooks.7 For the spherical cavity, s = 1/3 in any axis; for

11.2 Dielectric solvent effects

681

a thin slab cavity, s = 1 in the normal direction and s = 0 in plane; and for a long cylindrical cavity, s = 0 in the longitudinal axis and s = 1/2 in the transverse direction. The shape of the ellipsoidal cavity is supposed to be primarily determined by the shape of the solute molecule. Typical cases are long polyenes and large planar aromatic hydrocarbons. One can assume s = 0 for the strong π → π * absorption bands of these molecules. For smaller molecules, however, one should assume s ≈ 1/3 regardless of the shape of the solute molecule, as the cavity shape then may be primarily determined by the solvent molecules rather than Figure 11.2.1. Dependence of f ′ / f on the refractive in- the solute molecule. Note that Eq. [11.2.2] dex n for different values of s. [After reference 6] gives the Chako formula for s = 1/3, i.e., for the spherical cavity. For transitions whose moments are in the longitudinal axis of a long cylindrical cavity or in the plain of a thin slab cavity, Eq. [11.2.2] with s = 0 leads to f ′′ / f = 1/ n, so that the absorption intensity always decreases with the refractive index. If the transition moment is normal to a thin slab cavity, Eq. [11.2.2] with s = 1 leads to f ′′ / f = n 3 . The dependence of the ratio f ′′/f on the refractive index n according to Eq. [11.2.2] is illustrated for different values of s in Figure 11.2.1. The slope of the ratio is always positive for s > 1/4. For 0 < s < 1/4, it is negative in the region 1 ≤ n ≤ (1− s) / 3s and positive in the other region. Eq. [11.2.2] can be also written as the following form:

(nf ′′) =

(

)

fs n 2 − 1 + f

[11.2.3]

This equation shows a linear relationship between (nf ′′) and (n2 - 1). If a set of measured values of f ′′ vs. n are provided for a solute, the least-squares fitting to Eq. [11.2.3] of (nf ′′) against (n2 - 1) gives the values of f and s for the solute molecule. Note that f ′′ and f in Eq. [11.2.3] can be replaced by any quantities proportional to the oscillator strengths. Thus, they can be replaced by the integrated intensities or by their relative quantities. Figure 11.2.2 shows such plots for the π → π * absorption bands of β-carotene and the n → π * absorption bands of pyrazine measured8 in various organic solvents. Here, the relative intensities f ′′ / f c′′, where f c′′ is the absorption intensity measured in cyclohexane as the reference solvent, are considered, and y = nf ′′ / f c′′ is plotted against x = n2 - 1. The least-squares fittings give s = 0 for the allowed π → π * transition of β-carotene and s = 0.29 for the vibronic n → π * transition of pyrazine. Note that in this case the least-squares fitted line gives f / f c′′ as its intercept and s f / f c′′as its slope so that s is given as the ratio of the slope divided by the intercept. A similar study was made9 on the n → π * absorption bands of acetone and cyclopentanone, giving the results s = 0.88 and f = 1.8×10-4 for acetone and s=0.72 and f = 2.2×10-4 for cyclopentanone. A similar generalization was also made by Buckingham.10 He followed Kirkwood’s 11 idea in deriving the electric moment of a dielectric specimen produced by a fixed mole-

682

Tai-ichi Shibuya

cule in its interior and Scholte’s extension12 of the cavity field and the reaction field in the Onsager-BØttcher theory13,14 to an ellipsoidal cavity. Buckingham’s formula involves the polarizability of the solute molecule and appears quite different from Eq. [11.2.2]. It was shown6 that the Buckingham formula reduces to Eq. [11.2.2]. 11.2.3 DIELECTRIC SOLVENT Figure 11.2.2. Plots of y = nf ′ / f c′ vs. x = n2 - 1 for the π → π * EFFECT ON THE absorption bands of β-carotene (crosses) and the n → π * abRADIATIVE RATE sorption bands of pyrazine (solid circles). [After reference 6] CONSTANT The radiative rate constant is related to the absorption intensity of the transition from the ground state to the excited state under consideration. The application of Eq. [11.2.2] leads15 to

[(

) ]

k ′′r / k r = n s n 2 − 1 + 1

2

[11.2.4]

where: k r′ kr

apparent radiative rate constant of the solute molecule measured in a solvent of the refractive index n radiative rate constant of the molecule in its isolated state

Note that the local-field correction factor n[s(n2 - 1) + 1]2 varies from n to n5 as s varies from 0 to 1. For 9,10-diphenylanthracene (DPA), the correction factor was given15 as n[(0.128)(n2 - 1) +1]2, which lies between n and n2. This agrees with the observed data16 of fluorescence lifetimes of DPA in various solvents. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

N. Q. Chako, J. Chem. Phys., 2, 644 (1934). G. Kortòm, Z. Phys. Chem., B33, 243 (1936). (a) V. Henri and L. W. Pickett, J. Chem. Phys., 7, 439 (1939); (b) L. W. Pickett, E. Paddock, and E. Sackter, J. Am. Chem. Soc., 63, 1073 (1941). L. E. Jacobs and J. R. Platt, J. Chem. Phys., 16, 1137 (1948). S. R. Polo and M. K. Wilson, J. Chem. Phys., 23, 2376 (1955). T. Shibuya, J. Chem. Phys., 78, 5176 (1983). C. Kittel, Introduction to Solid State Physics, 4th Ed., Wiley, New York, 1971, Chap. 13. A. B. Myers and R. R. Birge, J. Chem. Phys., 73, 5314 (1980). T. Shibuya, Bull. Chem. Soc. Jpn. , 57, 2991 (1984). A. D. Buckingham, Proc. Roy. Soc. (London), A248, 169 (1958); A255, 32 (1960). J. G. Kirkwood, J. Chem. Phys., 7, 911 (1939). T. G. Scholte, Physica (Utrecht), 15, 437 (1949). L. Onsager, J. Am. Chem. Soc., 58, 1486 (1936). C. J. F. BØttcher, (a) Physica (Utrecht), 9, 937, 945 (1942); (b) Theory of Electric Polarization, Elsevier, New York, 1952; 2nd Ed., 1973, Vol. I. T. Shibuya, Chem. Phys. Lett., 103, 46 (1983). R. A. Lampert, S. R. Meech, J. Metcalfe, D. Phillips, A. P. Schaap, Chem. Phys. Lett., 94, 137 (1983).

12

Other Properties of Solvents, Solutions, and Products Obtained from Solutions 12.1 RHEOLOGICAL PROPERTIES, AGGREGATION, PERMEABILITY, MOLECULAR STRUCTURE, CRYSTALLINITY, AND OTHER PROPERTIES AFFECTED BY SOLVENTS George Wypych ChemTec Laboratories, Inc., Toronto, Canada

12.1.1 RHEOLOGICAL PROPERTIES The modification of rheological properties is one of the main reasons for adding solvents to various formulations. Rheology is also a separate complex subject which requires an in-depth understanding that can only be accomplished by consulting specialized sources such as monographic books on rheology fundamentals.1-3 Rheology is such a vast subject that the following discussion will only outline some of the important effects of solvents. When considering the viscosity of solvent mixtures, solvents can be divided into two groups: interacting and non-interacting solvents. The viscosity of a mixture of non-interacting solvents can be predicted with good approximation by a simple additive rule rule: i =n

log η = ∑ φi log η i

[12.1.1]

i =1

where: η i φ ηi

viscosity of solvent mixture iteration subscript for mixture components (i = 1, 2, 3, ..., n) fraction of component i viscosity of component i.

Interacting solvents contain either strong polar solvents or solvents which have the ability to form hydrogen bonds or influence each other on the basis of acid-base interaction. Solvent mixtures are complicated because of the changes in interaction that occurs with changes in the concentration of the components. Some general relationships describe vis-

684

George Wypych

cosity of such mixtures but none is sufficiently universal to replace measurement. Further details on solvent mixtures are included in Chapter 9. The addition of solute(s) further complicates rheology because in such mixtures solvents may not only interact among themselves but also with the solute(s). There are also interactions between solutes and the effect of ionized species with and without solvent participation. Only very dilute solutions of low molecular weight substances exhibit Newtonian viscosity. In these solutions, viscosity is a constant, proportionality factor of shear rate and shear stress. The viscosity of these solutions is usually well described by the classical, Einstein’s equation: η = ηs (1 + 2.5φ)

[12.1.2]

where: ηs φ

solvent viscosity volume fraction of spheres (e.g. suspended filler) or polymer fraction

If φ is expressed in solute mass concentration, the following relationship is used: φ=

NVc M

[12.1.3]

where: N V c M

Avogadro’s number molecular volume of solute ((4/3)πR3) with R - radius solute mass concentration molecular weight

Combination of equations [12.1.2] and [12.1.3] gives: η − ηs 2.5NV = ηs c M

[12.1.4]

The results of studies of polymer solutions are most frequently expressed in terms of intrinsic, specific, and relative viscosities and radius of gyration; the mathematical meaning of these and the relationships between them are given below:  η − ηs [η] = lim  c→ 0  ηs c ηsp c

  

= [η] + k1 [η] 2 c + ⋅ ⋅ ⋅

ln η r = [η] − k1′ [η] 2 c + ⋅ ⋅ ⋅ c η r = ηsp + 1 =

η ηs

[12.1.5]

[12.1.6]

[12.1.7]

[12.1.8]

12.1 Rheological properties, aggregation, permeability

1000

685

1.2

Log viscosity, cP

Viscosity, Pa s

1 100

10

0.8 0.6 0.4

1 0.001 0.01

0.1

1

Shear rate, s-1

10

100

0.2

-1

-0.8 -0.6 -0.4 -0.2

0

0.2

-1

Log concentration, g dl

Figure 12.1.1. Viscosity vs. shear rate for 10% solution Figure 12.1.2. Viscosity of polyphenylene solution in pyrilidinone. [Data from F. Motamedi, M Isomaki, of polyisobutylene in pristane. [Data from C R Schultheisz, G B McKenna, Antec ‘99, SPE, New M S Trimmer, Antec ‘98, SPE, Atlanta, 1998, p. 1772.] York, 1999, p 1125.] where: [η] ηsp ηr k1 k1′

intrinsic viscosity specific viscosity relative viscosity coefficient of direct interactions between pairs of molecules coefficient of indirect (hydrodynamic) interactions between pairs of molecules

In Θ solvents, the radius of gyration of unperturbed Gaussian chain enters the following relationship: [η] 0 =

Φ 0 Rg3, 0 M

[12.1.9]

where: Φ0 R g,0

coefficient of intramolecular hydrodynamic interactions = 3.16± 0.5×1024 radius of gyration of unperturbed Gaussian chain

In good solvents, the expansion of chains causes an increase of viscosity as described by the following equation: [η] =

Φ 0 α η3 Rg3, 0 M

[12.1.10]

where: αη

= [ η]1/3 / [ η]10/3 is and effective chain expansion factor.

Existing theories are far from being universal and precise in prediction of experimental data. A more complex treatment of measurement data is needed to obtain characteristics of these “rheological” liquids. Figure 12.1.1 shows that the viscosity of a solution depends on shear rate. These data comes from the development of a standard for instrument calibration by NIST to improve

686

George Wypych

109

500 400

7 -1

10

Shear rate, s

Viscosity, Poise

108

106 10

5

10

4

300 200 100

1000 100 1000

0 4

5

6

10 10 10 Molecular weight, Daltons

10

7

0

5

10

15

20

-6

Molecular weight x 10

Figure 12.1.3. Viscosity of 40% polystyrene in Figure 12.1.4. Shear rate of polystyrene in DOP vs. modi-2-ethyl hexyl phthalate. [Data from G D J Phillies, lecular weight. [Data from M Ponitsch, T Hollfelder, J Springer, Polym. Bull., 40, No.2-3, 345-52 (1998).] Macromolecules, 28, No.24, 8198-208 (1995).]

1.2

1.4

DXN

1.1

1.2 CCl

1

4

rel

0.9 0.8

log η

Viscosity, mPa s

1

DCE

0.7

0.6

associated unassociated

0.4

0.6

CHCl3

0.5 THF 0.4

0.8

TOL

0.2

DCM

0

0

10

20

30

40

50

PMMA concentration, % Figure 12.1.5. Viscosity of PMMA solutions in different solvents vs. PMMA concentration. Basic solvents: tetrahydrofuran, THF, and dioxane, DXN; neutral: toluene, TOL and CCl4; acidic: 1,2-dichloroethane, DCE, CHCl3, and dichloromethane, DCM. [Adapted, by permission, from M L Abel, M M Chehimi, Synthetic Metals, 66, No.3, 225-33 (1994).]

0

0.5

1

1.5

2

2.5

3

Concentration, g dl-1 Figure 12.1.6. Relative viscosity of block copolymers with and and without segments capable of forming complexes vs. concentration. [Data from I C De Witte, B G Bogdanov, E J Goethals, Macromol. Symp., 118, 237-46 (1997).]

the accuracy of measurements by application of nonlinear liquid standards.4 Figure 12.1.2 shows the effect of polymer concentration on the viscosity of a solution of polyphenylene in N-methyl pyrilidinone.5 Two regimes are clearly visible. The regimes are divided by a critical concentration above which viscosity increases more rapidly due to the interaction of chains leading to aggregate formation. These two sets of data show that there


400

80 MEK toluene

PEO/PAA PEO PAA

350 300

60

Viscosity, Poise

3

Intrinsic viscosity, cm g

-1

70

50 40 30 20 10

687

250 200 150 100 50

0

20

40

60

80

100

PVAc, wt% Figure 12.1.7. Intrinsic viscosity of PS/PVAc mixtures in methyl-ethyl-ketone, MEK, and toluene. [Data from H Raval, S Devi, Angew. Makromol. Chem., 227, 27-34 (1995).]

0

0

2

4

6

8

10

12

14

pH Figure 12.1.8. Viscosity of poly(ethylene oxide), PEO, poly(acrylic acid), PAA, and their 1:1 mixture in aqueous solution vs. pH. [Adapted, by permission, from I C De Witte, B G Bogdanov, E J Goethals, Macromol. Symp., 118, 237-46 (1997).

are considerable departures from the simple predictions of the above equations because, based on them, viscosity should be a simple function of molecular weight. Figure 12.1.3 shows, in addition, that the relationship between viscosity of the solution and molecular weight is nonlinear.6 Also, the critical shear rate, at which aggregates are formed, is a nonlinear function of molecular weight (Figure 12.1.4).7 These departures from simple relationships are representative of simple solutions. The relationships for viscosities of solution become even more complex if stronger interactions are included, such as the presence of different solvents, the presence of interacting groups within polymer, combinations of polymers, or the presence of electrostatic interactions between ionized structures within the same or different chains. Figure 12.1.5 gives one example of complex behavior of a polymer in solution. The viscosity of PMMA dissolved in different solvents depends on concentration but there is not one consistent relationship (Figure 12.1.5). Instead, three separate relationships exist each for basic, neutral, and acid solvents, respectively. This shows that solvent acid-base properties have a very strong influence on viscosity. Figure 12.1.6 shows two different behaviors for unassociated and associated block copolymers. The first type has a linear relationship between viscosity and concentration whereas with the second there is a rapid increase in viscosity as concentration increases. This is the best described as a power law function.8 Two polymers in combination have different reactions when dissolved in different solvents (Figure 12.1.7). In MEK, intrinsic viscosity increases as polymer concentration increases. In toluene, intrinsic viscosity decreases as polymer concentration increases.9 The polymer-solvent interaction term for MEK is very small (0.13) indicating a stable compatible system. The interaction term for toluene is much larger (0.58) which indicates a decreased compatibility of polymers in toluene and lowers viscosity of the mixture. Figure 12.1.8 explicitly shows that the behavior of

688

George Wypych

106

104 4 wt% CO

2

no CO2 Viscosity, Pa s

Viscosity, Poise

20% solvent no solvent

105

104

1

10

100 -1

Shear rate, s

1000

1000

100 10

100

1000 -1

Shear rate, s

Figure 12.1.9. Apparent melt viscosity of original PET Figure 12.1.10. Viscosity behavior of PS with and and PET containing 20% 1-methyl naphthalene vs. without CO2. [Data from M Lee, C Tzoganikis, C B Park, Antec ‘99, SPE, New York, 1999, p 2806.] shear rate. [Adapted, by permission, from S Tate, S Chiba, K Tani, Polymer, 37, No.19, 4421-4 (1996).]

individual polymers does not necessarily have a bearing on the viscosity of their solutions. Both poly(ethylene oxide) and poly(propylene oxide) are not affected by solution pH but, when used in combination, they become sensitive to solution pH. A rapid increase of viscosity at a lower pH is ascribed to intermolecular complex formation. This behavior can be used for thickening of formulations.8 Figures 12.1.9 and 12.1.10 show one potential application in which a small quantity of solvents can be used to lower melt viscosity during polymer processing. Figure 12.1.9 shows that not only can melt viscosity be reduced but also that the viscosity is almost independent of shear rate.10 In environmentally friendly process supercritical fluids can be used to reduce melt viscosity. The above data illustrate that the real behavior of solutions is much more complex than it is intuitively predicted based on simple models and relationships. The proper selection of solvent can be used to tailor the properties of formulation to the processing and application needs. Solution viscosity can be either increased or decreased to meet process technology requirements or to give the desired material properties.

REFERENCES 1 2 3 4 5 6 7 8 9

A Ya Malkin, Rheology Fundamentals, ChemTec Publishing, Toronto, 1994. Ch W Macosko, Rheology. Principles, Measurements, and Applications, VCH Publishers, New York, 1994. R I Tanner, K. Walters, Rheology: an Historical Perspective, Elsevier, Amsterdam, 1998. C R Schultheisz, G B McKenna, Antec ‘99, SPE, New York, 1999, p 1125. F Motamedi, M Isomaki, M S Trimmer, Antec ‘98, SPE, Atlanta, 1998, p. 1772. G D J Phillies, Macromolecules, 28, No.24, 8198-208 (1995). M Ponitsch, T Hollfelder, J Springer, Polym. Bull., 40, No.2-3, 345-52 (1998). I C De Witte, B G Bogdanov, E J Goethals, Macromol. Symp., 118, 237-46 (1997). H Raval, S Devi, Angew. Makromol. Chem., 227, 27-34 (1995).


10 11 12

689

S Tate, S Chiba, K Tani, Polymer, 37, No.19, 4421-4 (1996). M L Abel, M M Chehimi, Synthetic Metals, 66, No.3, 225-33 (1994). M Lee, C Tzoganikis, C B Park, Antec ‘99, SPE, New York, 1999, p 2806.

12.1.2 AGGREGATION The development of materials with an engineered morphological structure, such as selective membranes and nanostructures, employs principles of aggregation in these interesting technical solutions. Here, we consider some basic principles of aggregation, methods of studies, and outcomes. The discipline is relatively new therefore for the most part, only exploratory findings are available now. The theoretical understanding is still to be developed and this development is essential for the control of industrial processes and development of new materials. Methods of study and data interpretation still require further work and refinement. Several experimental techniques are used, including: microscopy (TEM, SEM);1,2 dynamic light scattering3-6 using laser sources, goniometers, and digital correlators; spectroscopic methods (UV, CD, fluorescence);7,8 fractionation; solubility and viscosity measurements;9 and acid-base interaction.10 Dynamic light scattering is the most popular method. Results are usually expressed by the radius of gyration, Rg, the second viral coefficient, A2, the association number, p, and the number of arms, f, for starlike micelles. Zimm’s plot and equation permits Rg and A2 to be estimated: 2  1  16 2 Rg KC = π 2 sin 2 θ + ⋅ ⋅ ⋅ + 2A2 C + ⋅ ⋅ ⋅ 1 + 3 Rθ Mw  λ0 

[12.1.11]

where: K C Rθ Mw λ0 θ

optical constant polymer concentration Rayleigh ratio for the solution weight average molecular weight wavelength of light scattering angle.

The following equations are used to calculate p and f: M agg

p=

M1

Rg Rgarm

; M agg = KI q→0 / C

= f (1−v )/ 2

where: Magg M1 I Rgarm v

mass of aggregates mass of free copolymers scattered intensity radius of gyration of linear polymer excluded volume exponent.

[12.1.12]

[12.1.13]

690

George Wypych

Figure 12.1.11. Morphological features: starlike and crew-cut. The bottom drawing illustrates polymer brush. [Adapted, by permission from G Liu, Macromol. Symp., 113, 233 (1997).]

Figure 12.1.12. Morphological features of PS-PAA copolymer having different ratios of PS and PAA block lengths: A - 8.3, B- 12, C - 20.5, and D - 50. [Adapted, by permission, from L Zhang, A Eisenberg, Macromol. Symp., 113, 221-32 (1997).]

Figure 12.1.13. PS-PAA aggregates of different morphologies depending on its concentration in DMF: A - 2, B - 2.6, C - 3, D - 4 wt%. [Adapted, by permission, from L Zhang, A Eisenberg, Macromol. Symp., 113, 221-32 (1997).]

Figure 12.1.14. TEM micrograph of nanospheres. [Adapted, by permission, from G Liu, Macromol. Symp., 113, 233 (1997).]

Three major morphological features are under investigation: starlike, crew-cut, and polymer brushes (Figure 12.1.11). Morphological features have been given nick-names characterizing the observed shapes, such as “animals” or “flowers” to distinguish between various observed images.6 Figure 12.1.12 shows four morphologies of aggregates formed by polystyrene, PS,-poly(acrylic acid), PAA, diblock copolymers.1 The morphology produced was a direct result of the ratio between lengths of blocks of PS and PAA (see Figure 12.1.12). When this ratio is low (8.3), spherical micelles are formed. With a slightly higher ratio (12), rod-like micelles result which have narrow distribution of diameter but variable length. Increasing ratio even further (20.5) causes vesicular aggregates to form. With the highest ratio (50), large spherical micelles are formed. One of the reasons for the differences in these formations is that the surface tension between the core and the solvent varies widely. In order to decrease interfacial tension between the core and the solvent, the aggregate increases the core size.1 Polymer concentration in solution also affects aggregate for-


Figure 12.1.15. TEM micrograph of assemblies of cylindrical aggregates. [Adapted, by permission, from G Liu, Macromol. Symp., 113, 233 (1997).]

691

Figure 12.1.16. TEM micrograph of nanofibers. [Adapted, by permission, from G Liu, Macromol. Symp., 113, 233 (1997).]

350

PVC/BrBz PVC/DOA

300

Rg, nm

250 200 150 100 50 20

30

40

50

60

70

o

Temperature, C Figure 12.1.17. TEM micrograph of knots and strands formed in carbohydrate amphiphile. [Adapted, by permission, from U Beginn, S Keinath, M Moller, Macromol. Chem. Phys., 199, No.11, 2379-84 (1998).]

Figure 12.1.18. Radius of gyration vs. temperature for PVC solutions in bromobenzene, BrBz, and dioxane, DOA. [Adapted, by permission, from Hong Po-Da, Chen Jean-Hong Chen, Polymer, 40, 4077-4085, (1999).]

mation as Figure 12.1.13 shows. As polymer concentration increases, spherical micelles are replaced by a mixture of rod-like structures and vesicles. With a further increase in concentration, the rod-like shapes disappear and only vesicles remain. Figures 12.1.14-12.1.16 show a variety of shapes which have been observed. These include nanospheres, assemblies of cylindrical aggregates, and nanofibers.11 Nanospheres were obtained by the gradual removal of solvent by dialysis, fibers were produced by a series of processes involving dissolution, crosslinking, and annealing. Figure 12.1.17 sheds some light on the mechanism of aggregate formation. Two elements are clearly visible from micrographs: knots and strands. Based on studies of carbohydrate amphiphiles, it is concluded that knots are formed early in the process by spinodal decomposition. Formation of

692

George Wypych

100

Association number, p

1000

Specific viscosity

100

10

1

0.1

1

10 -1

Concentration, g l

Figure 12.1.19. Specific viscosity of PVC in bromobenzene solution vs. concentration. [Adapted, by permission, from Hong Po-Da, Chen Jean-Hong Chen, Polymer, 40, 4077-4085, (1999).]

1

10

1 0.001

0.01

0.1

1 -3

Concentration, mg cm

Figure 12.1.20. Diblock copolymer association number vs. its concentration. [Adapted, by permission, from D Lairez, M Adam, J-P Carton, E Raspaud, Macromolecules, 30, No.22, 6798-809 (1997).]

I560/I524

strands between knots occurs much later.2 A similar course of events occurs with the molecular aggregation of PVC solutions.5 Figure 12.1.18 gives more details on the mechanism of aggregate formation.5 PVC dissolved in bromobenzene, BrBz, 0.1 undergoes a coil to globule transition which does not occur in dioxane, DOA, solution. Bromobenzene has a larger molar volume than dioxane and the polymer chain must readjust to the solvent molar volume in order to interact. Figure 12.1.19 0.01 0.01 0.1 1 10 100 shows that two regimes of aggregation are involved which are divided by a certain Concentration, g dl-1 critical value. Below the critical value Figure 12.1.21. Fluorescence intensity ratio vs. concenchains are far apart and do not interact due tration of polyimide in chloroform. [Data from H Luo, L Dong, H Tang, F Teng, Z Feng, Macromol. Chem. to contact self-avoidance. At the critical Phys., 200, No.3, 629-34 (1999).] point, knots, similar to shown in Figure 12.1.17, begin to form and the resultant aggregation increases viscosity. As the polymer concentration increases, the association number, p, also increases, but then levels off (Figure 12.1.20). Fluorescence studies (Figure 12.1.21) show that there are two characteristic points relative to concentration. Below the first point, at 0.13 g/dl, in very dilute solution, molecules are highly expanded and distant from one another and fluorescence does not change. Between the two points, individual chain coils begin to sense each other and become affected


693

by the presence of neighbors forming intermolecular associations. Above the second point, coils begin to overlap leading to dense packing.8 These data are essentially similar to those presented in Figure 12.1.20 but cover a broader concentration range. The above studies show that there are many means of regulating aggregate size and shape which is likely to become an essential method of modifying materials by morphological engineering.

REFERENCES 1 2 3 4 5 6 7 8 9 10 11

L Zhang, A Eisenberg, Macromol. Symp., 113, 221-32 (1997). U Beginn, S Keinath, M Moller, Macromol. Chem. Phys., 199, No.11, 2379-84 (1998). P A Cirkel, T Okada, S Kinugasa, Macromolecules, 32, No.2, 531-3 (1999). K Chakrabarty, R A Weiss, A Sehgal, T A P Seery, Macromolecules, 31, No.21, 7390-7 (1998). Hong Po-Da, Chen Jean-Hong Chen, Polymer, 40, 4077-4085, (1999). D Lairez, M Adam, J-P Carton, E Raspaud, Macromolecules, 30, No.22, 6798-809 (1997). R Fiesel, C E Halkyard, M E Rampey, L Kloppenburg, S L Studer-Martinez, U Scherf, U H F Bunz, Macromol. Rapid Commun., 20, No.3, 107-11 (1999). H Luo, L Dong, H Tang, F Teng, Z Feng, Macromol. Chem. Phys., 200, No.3, 629-34 (1999). A Leiva, L Gargallo, D Radic, J. Macromol. Sci. B, 37, No.1, 45-57 (1998). S Bistac, J Schultz, Macromol. Chem. Phys., 198, No. 2, 531-5 (1997). G Liu, Macromol. Symp., 113, 233 (1997).

12.1.3 PERMEABILITY The phenomenon of solvent transport through solid barriers has three aspects which discussed under the heading of permeability. These are the permeation of solvent through materials (films, containers, etc.); the use of pervaporation membranes to separate organic solvents from water or water from solvents; the manufacture of permeate selective membranes. The permeability of different polymers and plastics to various solvents can be found in an extensive, specialized database published as a book and as a CD-ROM.1 The intrinsic properties of polymers can be modified in several ways to increase their resistance and reduce their permeability to solvents. The development of plastic gas tanks was a major driving force behind these developments and now various plastic containers are manufactured using similar processes. Fluorination of plastics, usually polypropylene or polyethylene, is by far the most common modification. Containers are typically manufactured by a blow molding process where they are protected by a fluorination process applied on line or, more frequently, off line. The first patent2 for this process was issued in 1975 and numerous other patents, some of them issued quite recently have made further improvements to the process.3 The latest processes use a reactive gas containing 0.1 to 1% fluorine to treat parison within the mold after it was expanded.3 Containers treated by this process are barrier to polar liquids, hydrocarbon fuels, and carbon fuels containing polar liquids such as alcohols, ethers, amines, carboxylic acids, ketones, etc. Superior performance is achieved when a hydrogen purge precedes the exposure of the container to oxygen. In another recent development,4 containers are being produced from a blend of polyethylene and poly(vinylidene fluoride) with aluminum stearate as compatibilizer. This process eliminates the use of the toxic gas fluorine and by-products which reduces environmental pollution and disposal problems. It is argued that fluorinated containers may not perform effectively when the protective layer becomes damaged due to stress cracking.

694

George Wypych

Such a blend is intended as a replacement for previous blends of polyamide and polyethylene used in gas tanks and other containers. These older blends were exposed to a mixture of hydrocarbons but now oxygenated solvents (e.g., methanol) have been added which cause an unacceptable reduction of barrier properties. A replacement technology is needed. Improved barrier properties in this design4 comes from increased crystallinity of the material brought about by controlling laminar thickness of polyethylene crystals and processing material under conditions which favor the crystallization of poly(vinylidene fluoride). Another approach involves the use of difunctional telechelic polyolefins with ester, hydroxyl, and amine terminal groups.5 Telechelic polyolefins can be used to make polyesters, polyamides, and polyurethanes with low permeability to solvents and gases. Separation processes such as ultrafiltration and microfiltration use porous membranes which allow the passage of molecules smaller than the membrane pore size. Ultrafiltration membranes have pore sizes from 0.001 to 0.1 µm while microfiltration membranes have pore sizes in the range of 0.02 to 10 µm. The production of these membranes is almost exclusively based on non-solvent inversion method which has two essential steps: the polymer is dissolved in a solvent, cast to form a film then the film is exposed to a non-solvent. Two factors determine the quality of the membrane: pore size and selectivity. Selectivity is determined by how narrow the distribution of pore size is.6,7 In order to obtain membranes with good selectivity, one must control the non-solvent inversion process so that it inverts slowly. If it occurs too fast, it causes the formation of pores of different sizes which will be non-uniformly distributed. This can be prevented either by an introduction of a large number of nuclei, which are uniformly distributed in the polymer membrane or by the use of a solvent combination which regulates the rate of solvent replacement. Typical solvents used in membrane production include: N-methylpyrrolidinone, N,N-dimethylacetamide, N,N-dimethylformamide, dimethylsulfoxide, tetrahydrofuran, dioxane, dichloromethane, methyl acetate, ethyl acetate, and chloroform. They are used alone or in mixtures.6 These are used most frequently as non-solvents: methanol, ethanol, 1-propanol, 2-propanol, 1-butanol, 2-butanol, and t-butanol. Polymers involved include: polysulfone, polyethersulfone, polyamide, polyimide, polyetherimide, polyolefins, polycarbonate, polyphenyleneoxide, poly(vinylidene fluoride), polyacrylonitrile, and cellulose and its derivatives. A second method of membrane preparation is based on the thermally-induced phase separation process.8 The goal of this process is to produce a membrane which has an ultrathin separation layer (to improve permeation flux) and uniform pores.9 In this process, polymer is usually dissolved in a mixture of solvents which allow the mixture to be processed either by spinning Figure 12.1.23. PEI fiber from or by coating. The desired Figure 12.1.22. PEI fiber from DMF. [Adapted, by permission membrane morphology is obNMP. [Adapted, by permission from D Wang, K Li, W K Teo, from D Wang, K Li, W K Teo, J. Appl. Polym. Sci., 71, No.11, tained through cooling to inJ. Appl. Polym. Sci., 71, No.11, duce phase separation of 1789-96 (1999).] 1789-96 (1999).]

695

40

1.6

35

1.4

30 25 20 15

propionic acid 2-propanol water

10 5 15 20 25 30 35 40 45 50 55 o

Temperature, C Figure 12.1.24. Precipitation value for PEI/NMP system with different non-solvents vs. temperature. [Data from D Wang, Li K, W K Teo, J. Appl. Polym. Sci., 71, No.11, 1789-96 (1999).]

Skin layer thickness,µm

Precipitation value, g


1.2 1 0.8 0.6 0.4 0.2

0

2

4

6

8

10

Evaporation time, min Figure 12.1.25. Skin thickness vs. evaporation time of formation of asymmetric membrane from polysulfone. [Data from A Yamasaki, R K Tyagi, A E Fouda, T Matsura, K Jonasson, J. Appl. Polym. Sci., 71, No.9, 1367-74 (1999).]

polymer/solvent/non-solvent system. One practices in use is the addition of non-solvents to the casting solution.10 The mixture should be designed such that homogeneous casting is still possible but the thermodynamic condition approaches phase separation. Because of cooling and/or evaporation of one of the solvents, demixing occurs before material enters coagulation bath.10 Figures 12.1.22 and 12.1.23 explain technical principles behind formation of efficient and selective membrane. Figure 12.1.22 shows a micrograph of hollow PEI fiber produced from N-methyl-2-pyrrolidone, NMP, which has thin surface layer and uniform pores and Figure 12.1.23 shows the same fiber obtained from a solution in dimethylformamide, DMF, which has a thick surface layer and less uniform pores.9 The effect depends on the interaction of polar and non-polar components. The compatibility of components was estimated based on their Hansen’s solubility parameter difference. The compatibility increases as the solubility parameter difference decreases.9 Adjusting temperature is another method of control because the Hansen’s solubility parameter decreases as the temperature increases. A procedure was developed to determine precipitation values by titration with non-solvent to a cloud point.9 Use of this procedure aids in selecting a suitable non-solvent for a given polymer/solvent system. Figure 12.1.24 shows the results from this method.9 Successful in membrane production by either non-solvent inversion or thermally-induced phase separation requires careful analysis of the compatibilities between polymer and solvent, polymer and non-solvent, and solvent and non-solvent. Also the processing regime, which includes temperature control, removal of volatile components, uniformity of solvent replacement must be carefully controlled. Efforts must be made to select solvents of low toxicity and to minimize solvent consumption. An older method of producing polyetherimide membranes involved the use of a mixture of two solvents: dichloromethane (very volatile) and 1,1,2-trichloroethane together

George Wypych

1000

35 30 25

-1 -2

20 15 10

-1

Separation factor

A/l B/l

-1

-2

A/l, mol s m Pa or B/l, mol s m Pa

-2

696

800

600

400

5 0

0

50

100

150

200

200

250

0

Evaporation time, min

0.8

1

100 Hexane in permeate, wt%

12

-1

10 -2

0.6

Figure 12.1.27. Separation factor vs. solubility of solvents in water. [Adapted, by permission, from M Hoshi, M Kobayashi, T Saitoh, A Higuchi, N Tsutomu, J. Appl. Polym. Sci., 69, No.8, 1483-94 (1998).]

14

Flux, g m h

0.4

Solubility of solvent in water, wt%

Figure 12.1.26. Pervaporation transport parameters: A/l (liquid transport) and B/l (vapor transport) vs. evaporation time during membrane preparation from aromatic polyamide. [Data from A Yamasaki, R K Tyagi, A Fouda, T Matsuura, J. Appl. Polym. Sci., 57, No.12, 1473-81 (1995).]

8 6 4 2 0

0.2

0

0.02

0.04

0.06

0.08

0.1

Triclene in feed, wt% Figure 12.1.28. Effect of triclene concentration in feed solution on its pervaporation flux through acrylic membrane. [Data from M Hoshi, M Kobayashi, T Saitoh, A Higuchi, N Tsutomu, J. Appl. Polym. Sci., 69, No.8, 1483-94 (1998).]

80 60 40 20 0

0

20

40

60

80

100

Hexane in feed, wt% Figure 12.1.29. Effect of hexane concentration in feed solution on its concentration in permeate through acrylic membrane. [Data from M Hoshi, M Kobayashi, T Saitoh, A Higuchi, N Tsutomu, J. Appl. Polym. Sci., 69, No.8, 1483-94 (1998).]

with a mixture of non-solvents: xylene and acetic acid. In addition, large amounts of acetone were used in the coagulation bath. A more recent process avoided environmental problems by using tetrahydrofuran and γ-butyrolactone in a process in which water only is used in the coagulating bath (membrane quality remained the same).10


697

Thermally-induced phase separation has been applied to the production of polysilane foams.11 Variation of polymer concentration, solvent type, and cooling rate have been used to refine the macrostructure. Both these membrane production methods rely on the formation of aggregates of controlled size and shape as discussed in the previous section. Figure 12.1.25 shows the effect of of solvent evaporation time on the skin layer thickness. It is only in the beginning of the process that skin thickness does not increase. Thereafter there is rapid skin thickness growth. Polymer concentration is the other important determinant of skin thickness. Figure 12.1.26 shows that during very short evaporation times (5 min), even though skin is thin the pores diameter is so small that the pervaporation parameters A/l and B/l, which characterize liquid and vapor transport, respectively, do not increase. Longer evaporation times bring about a gradual decrease in transport properties.12,13 Figure 12.1.27 shows the relationship between the separation factor and the solubility of solvents in water. The separation of solvent by a pervaporation membrane occurs less efficiently as solvent solubility increases.The more concentrated the solution of solvent, the faster is the separation (Figure 12.1.28). Separation of hexane from a mixture with heptane is similar (Figure 12.1.29). The acrylic membrane shows good selectivity. These examples demonstrate the usefulness of pervaporation membranes in solvent recovery processes.

REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Permeability and Other Film Properties, Plastics Design Library, Norwich, 1996. US Patent 3,862,284, 1975. J P Hobbs, J F DeiTos, M Anand, US Patent 5,770,135, Air Products and Chemicals, Inc., 1998 and US Patent 5,244,615, 1992. R T Robichaud, US Patent 5,702,786, Greif Bros. Corporation, 1997. P O Nubel, H B Yokelson, US Patent 5,731,383, Amoco Corporation, 1998 and US Patent 5,589,548, 1996. J M Hong, S R Ha, H C Park, Y S Kang, K H Ahn, US Patent 5,708,040, Korea Institute of Science and Technology, 1998. K-H Lee, J-G Jegal, Y-I Park, US Patent 5,868,975, Korea Institute of Science and Technology, 1999. J M Radovich, M Rothberg, G Washington, US Patent 5,645,778, Althin Medical, Inc., 1997. D Wang, Li K, W K Teo, J. Appl. Polym. Sci., 71, No.11, 1789-96 (1999). K-V Peinemann, J F Maggioni, S P Nunes, Polymer, 39, No.15, 3411-6 (1998). L L Whinnery, W R Even, J V Beach, D A Loy, J. Polym. Sci.: Polym. Chem. Ed., 34, No.8, 1623-7 (1996). A Yamasaki, R K Tyagi, A Fouda, T Matsuura, J. Appl. Polym. Sci., 57, No.12, 1473-81 (1995). A Yamasaki, R K Tyagi, A E Fouda, T Matsura, K Jonasson, J. Appl. Polym. Sci., 71, No.9, 1367-74 (1999). M Hoshi, M Kobayashi, T Saitoh, A Higuchi, N Tsutomu, J. Appl. Polym. Sci., 69, No.8, 1483-94 (1998).

12.1.4 MOLECULAR STRUCTURE AND CRYSTALLINITY The gelation of polymer-solvent systems was initially thought of as a process which occurs only in poor solvents which promote chain-chain aggregation. Further studies have revealed that many polymers also form gels in good solvents. This has prompted research which attempts to understand the mechanisms of gelation. These efforts have contributed the current understanding of the association between molecules of solvents and polymers.1 Studies on isotactic, iPS, and syndiotactic, sPS, polystyrenes2-6 have confirmed that, although both polymers have the same monomeric units, they are significantly different in terms of their solubility, gelation, and crystallization. In systems where sPS has been dissolved in benzene and carbon tetrachloride, gelation is accomplished in a few minutes

698

George Wypych

whereas in chloroform it takes dichloromethane tens of hours. no solvent 15 Adding a small amount of benzene to the chlo10 roform accelerates the gelation process. There is 5 also evidence from IR which shows that the 0 orientation of the 0 200 400 600 ring plane is perStrain, % pendicular to the Figure 12.1.30. Molecular Figure 12.1.31. Stress-strain behavior at 130oC of chain axis. This model of polystyrene and glassy sPS with and without immersion in that benzene rings alignment. dichloromethane. [Data from C Daniel, L Guadagno, suggests there is a strong [Adapted by permission, from V Vittoria, Macromol. Symp., 114, 217 (1997).] J M Guenet, Macromol. Symp., interaction be114, 97 (1997).] tween benzene and sPS molecules. The other interesting observation came 1715 from studies on decalin sPS and iPS systems. iPS forms a transparent gel and then becomes 1710 turbid, gradually forming trigonal crystallites of iPS. sPS does not convert to a gel but the 1705 fine crystalline precipitate particles instead. 1700 This shows that the molecular arrangement cyclohexanone depends on both the solvent type and on the N-methyl-2-pyrrolidone 1695 molecular structure of the polymer.2 Figure 12.1.30 shows how benzene mol1690 ecules align themselves parallel to the phenyl rings of polystyrene and how they are housed 1685 0 0.2 0.4 0.6 0.8 1 within the helical form of the polymer structure which is stabilized by the presence of solSolvent molar fraction vent.1 Work on poly(ethylene oxide) gels7 indiFigure 12.1.32. Carbonyl frequency vs. molar fraction of solvent. [Adapted, by permission, from cates that the presence of solvents such as J M Gomez-Elvira, P Tiemblo, G Martinez, J Milan, chloroform and carbon disulfide contributes to Macromol. Symp., 114, 151 (1997).] the formation of a uniform helical conformation. The gelation behavior and the gel structure depend on the solvent type which, in turn, is determined by solvent-polymer interaction. In a good solvent, polythiophene molecules exist in coiled conformation. In a poor solvent, the molecules form aggregates through the short substituents.8 Polyvinylchloride, PVC, which has a low crystallinity, gives strong gels. Neutron diffraction and scattering studies show that these strong gels result from the formation of Carbonyl frequency shift, cm

-1

Stress, MPa

20


699

Figure 12.1.33. Crystal structure of poly(ethylene oxide) molecular complex with: (a) p-dichlorobenzene, (b) resorcinol, (c) p-nitrophenol. [Adapted, by permission, from M Dosiere, Macromol. Symp., 114, 51 (1997).]

PVC-solvent complexes.9 This indicates that the presence of solvent may affect the mechanical properties of such a system. Figure 12.1.31 shows that the presence of dichloromethane in sPS changes the mechanical characteristic of the material. A solvent-free polymer has a high elongation and yield value. An oriented polymer containing dichloromethane has lower elongation, no yield value, and approximately four times greater tensile strength.10 The interaction of polymer-solvent affects rheological properties. Studies on a divinyl ether-maleic anhydride copolymer show that molecular structure of the copolymer can be altered by the solvent selected for synthesis.11 Studies Figure 12.1.34. Optical micrograph stability of PVC is afof PEO-resorcinol complex. have shown that the thermal and UV 12-14 fected by the presence of solvents. [Adapted, by permission, from M Dosiere, Macromol. Symp., 114, Figure 12.1.32 shows that solvent type and its molar 51 (1997).] fraction affect the value of the carbonyl frequency shift. This frequency shift occurs at a very low solvent concentration, The slope angles for each solvent are noticeably different.15 Similar frequency shifts were reported for various solvents in polyetheretherketone, PEEK, solutions.16 These associations are precursors of further ordering by crystallization. Figure 12.1.33 shows different crystalline structures of poly(ethylene oxide) with various solvents.17 Depending on the solvent type, a unit cell contains a variable number of monomers and molecules of solvent stacked along the crystallographic axis. In the case of dichlorobenzene, the molecular complex is orthorhombic with 10 monomers and 3 solvent molecules. Figure 12.1.34 shows optical micrograph of spherulite formed from unit cells having 8 monomers and 4 molecules of solvent. The unit cell of the triclinic crystal, formed with nitrophenol, is composed of 6 monomers and 4 molecules of solvent. This underlines the importance of solvent selection in achieving the desired structure in the formed material. The technique is also advantageous in polymer synthesis, in improving polymer stability, etc.

700

George Wypych

Temperature is an essential parameter in the crystallization process.18 Rapid cooling of a polycarbonate, PC, solution in benzene resulted in extremely high crystallinity (46.4%) as compared to the typical PC crystallinity of about 30%. Polymer-solvent interaction combined with the application of an external force leads to the surface crazing of materials. The process is based on similar principles as discussed in this section formation of fibrilar crystalline structures. Although research on molecular structure and crystallization is yet to formulate a theoretical background which might predict the effect of different solvents on the fine structure of different polymers, studies have uncovered numerous issues which cause concern but which also point to new applications. A major concern is the effect of solvents on craze formation, and on thermal and UV degradation. Potential applications include engineering of polymer morphology by synthesis in the presence of selected solvent under a controlled thermal regime, polymer reinforcement by interaction with smaller molecules, better retention of additives, and modification of surface properties to change adhesion or to improve surface uniformity, etc.

REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

J M Guenet, Macromol. Symp., 114, 97 (1997). M Kobayashi, Macromol. Symp., 114, 1-12 (1997). T Nakaoki, M Kobayashi, J. Mol. Struct., 242, 315 (1991). M Kobayashi, T Nakoaki, Macromolecules, 23, 78 (1990). M Kobayashi, T. Kozasa, Appl. Spectrosc., 47, 1417 (1993). M. Kobayashi, T Yoshika, M Imai, Y Itoh, Macromolecules, 28, 7376 (1995). M Kobayashi, K Kitagawa, Macromol. Symp., 114, 291 (1997). P V Shibaev, K Schaumburg, T Bjornholm, K Norgaard, Synthetic Metals, 97, No.2, 97-104 (1998). H Reinecke, J M Guenet, C. Mijangos, Macromol. Symp., 114, 309 (1997). C Daniel, L Guadagno, V Vittoria, Macromol. Symp., 114, 217 (1997). M Y Gorshkova, T L Lebedeva, L L Stotskaya, I Y Slonim, Polym. Sci. Ser. A, 38, No.10, 1094-6 (1996). G Wypych, Handbook of Material Weathering, ChemTec Publishing, Toronto, 1995. G Wypych, Poly(vinyl chloride) degradation, Elsevier, Amsterdam, 1985. G Wypych, Poly(vinyl chloride) stabilization, Elsevier, Amsterdam, 1986. J M Gomez-Elvira, P Tiemblo, G Martinez, J Milan, Macromol. Symp., 114, 151 (1997). B H Stuart, D R Williams, Polymer, 36, No.22, 4209-13 (1995). M Dosiere, Macromol. Symp., 114, 51 (1997). Gending Ji, Fengting Li, Wei Zhu, Qingping Dai, Gi Xue, Xinhong Gu, J. Macromol. Sci. A, A34, No.2, 369-76 (1997). I V Bykova, E A Sinevich, S N Chvalun, N F Bakeev, Polym. Sci. Ser. A, 39, No.1, 105-12 (1997).

12.1.5 OTHER PROPERTIES AFFECTED BY SOLVENTS Many other properties of solutes and solutions are affected by solvents. Here, we will discuss material stability and stabilization, some aspects of reactivity (more information on this subject appears in Chapter 13), physical properties, some aspects of electrical and electrochemical properties (more information on this subject appears in Chapter 11), surface properties, and polarity and donor properties of solvents. Degradative processes have long been known to be promoted by the products of solvent degradation. Tetrahydrofuran is oxidized to form a peroxide which then dissociates to form two radicals initiating a chain of photo-oxidation reactions. Figure 12.1.35 shows the kinetics of hydroperoxide formation.1 Similar observations, but in polymer system, were made in xylene by direct determination of the radicals formed using ESR.2 An increased concentration in trace quantities of xylene contributed to the formation of n-octane radicals

0.5 0.4 0.3 0.2 0.1 0

0

0.5

1

1.5

2

2.5

Time, h Figure 12.1.35. Peroxide formation from tetrahydrofuran during irradiation at 254 nm. [Data from J F Rabek, T A Skowronski, B Ranby, Polymer, 21, 226 (1980).]

3

Relative signal intensity, arbitrary units

Peroxide concentration, arbitrary units


701

5 4 3 2 1 0 -5 10

0.0001 0.001 0.01 Xylene concentration, mol l-1

0.1

Figure 12.1.36. Spectral intensity of n-octane radical formed during irradiation of PE at -196oC for 30 min vs. concentration of p-xylene. [Adapted, by permission, from H Kubota, M Kimura, Polym. Deg. Stab., 38, 1 (1992).]

by abstracting hydrogen from polyethylene chain in an α-position to the double bond (Figure 12.1.36). 3-hydroperoxyhexane cleaves with formation of carboxylic acid and hydrocarbon radical (ethylene or propylene).3 These known examples show that the presence of even traces of solvents may change the chemistry and the rate of photo-oxidative processes because of formation of radicals. In another aspect of photo-oxidative processes, solvents influence the reactivity of small molecules and chain segments by facilitating the mobility of molecules and changing the absorption of light, the wavelength of emitted fluorescent radiation, and the lifetime of radicals. Work on anthraquinone derivatives, which are common photosensitizers, has shown that when photosensitizer is dissolved in isopropanol (hydrogen-donating solvent) the half-life of radicals is increased by a factor of seven compared to acetonitrile (a non-hydrogen-donating solvent).4 This shows that the ability of a photosensitizer to act in this manner depends on the presence of a hydrogen donor (frequently the solvent). In studies on another group of photosensitizers − 1,2-diketones, the solvent cyclohexane increased the absorption wavelength of the sensitizer in the UV range by more than 10 nm compared with solutions in ethanol and chloroform.5 Such a change in absorption may benefit some systems because the energy of absorbed radiation will become lower than the energy required to disrupt existing chemical bonds. But it may also increase the potential for degradation by shifting radiation wavelength to the range absorbed by a particular material. Presence of water may have a plasticizing action which increases the mobility of chains and their potential for interactions and reactions.6 Studies on photoresists, based on methacryloyethyl- phenylglyoxylate, show that, in aprotic solvents, the main reaction mechanism is a Norrish type II photolysis leading to chain scission.7 In aprotic solvents, the polymer is photoreduced and crosslinks are formed.

702

George Wypych 105

80 70

10

4

60 Yield, %

log η0

1000 100

50 40 30 20

10

10 1 4 10

5

10 log M

6

10

w

Figure 12.1.37. Effect of viscosity on molecular weight of polyethylene wax obtained by thermolysis in the presence of phenylether at 370oC. [Adapted, by permission, from L Guy, B Fixari, Polymer, 40, No.10, 2845-57 (1999).]

0

methanol

DMF

DMSO

Solvent Figure 12.1.38. Yield of condensation products depending on solvent. [Adapted, by permission, from J Jeczalik, J. Polym. Sci.: Polym. Chem. Ed., 34, No.6, 1083-5 (1996).]

Polarity of the solvent determines quantum yields in polyimides.8 The most efficient photocleavage was in medium-polar solvents. The selection of solvent may change the chemical mechanism of degradation and the associated products of such reactions. Singlet oxygen is known to affect material stability by its ability to react directly with hydrocarbon chains to form peroxides. The photosensitizers discussed above are capable of generating singlet oxygen. Solvents, in addition to their ability to promote photosensitizer action, affect the lifetime of singlet oxygen (the time it has to react with molecules and form peroxides). Singlet oxygen has very short lifetime in water (2 µs) but much longer times in various solvents (e.g., 24 µs in benzene, 200 µs in carbon disulfide, and 700 µs in carbon tetrachloride).9 Solvents also increase the oxygen diffusion coefficient in a polymer. It was calculated10 that in solid polystyrene only less than 2% of singlet oxygen is quenched compared with more than 50% in solution. Also, the quantum yield of singlet oxygen was only 0.56 in polystyrene and 0.83 in its benzene solution.10 The photostabilizer must be durable. It was found that salicylic stabilizers are efficiently degraded by singlet oxygen in polar alkaline media but in a less polar, non-alkaline solvents these stabilizers are durable.11 In hydrogen-bonding solvents, the absorption spectrum of UV absorbers is changed.12 Thermal decomposition yield and composition of polystyrene wastes is affected by addition of solvents.13 Solvents play two roles: First, they stabilize radicals and, second, by lowering solution viscosity they contribute to homogeneity, increase the mobility of components and increase reactivity. 85% benzyl and phenoxy radicals were stabilized by hydrogens from tetralin.13 In polyethylene thermolysis, the use of solvent to reduce viscosity resulted in obtaining lower molecular weight, narrow molecular weight distribution and high crystallinity in resultant polyethylene wax.14 Figure 12.1.37 illustrates the relationship


100

Surface DMS, molar %

Glass transition temperature, K

165

160

155

150

145

703

0

5

10

15

20

25

MIBK concentration, % Figure 12.1.39. Glass transition temperature of perfluoro polyoxyalkylene oligomers vs. MIBK concentration. [Data from S Turri, M Scicchitano, G Gianotti, C Tonelli, Eur. Polym. J., 31, No.12, 1227-33 (1995).]

98 96 2% DMS 1% DMS

94 92 90

0

5

10

15

20

25

30

Toluene content, v% Figure 12.11.40. Surface DMS vs. toluene concentration for two bulk contents of DMS in polystyrene. [Data from Jiaxing Chen, J A Gardella, Macromolecules, 31, No.26, 9328-36 (1998).]

between molecular weight and viscosity of solutions of polyethylene in phenylether as a solvent.14 Figure 12.1.38 shows that polymerization yield depends on the solvent selected for the polymerization reaction. The polymerization yield increases as solvent polarity increases.15 Work on copolymerization of methyl methacrylate and N-vinylpyrrolidone shows that the solvent selected regulates the composition of copolymer.16 The smaller the polarity of the solvent and the lower the difference between the resonance factors of the two monomers the more readily they can copolymerize. Work on electropolymerization17 has shown an extreme case of solvent effect in the electrografting of polymer on metal. If the donocity of the monomer is too high compared with that of the solvent, no electrografting occurs. If the donocities are low, a high dielectric constant of solvent decreases grafting efficiency. This work, which may be very important in corrosion protection, illustrates that a variety of influences may affect the polymerization reaction. The addition of solvent to polymer has a plasticizing effect. The increase in free volume has a further influence on the glass transition temperature of the polymer. Figure 12.1.39 shows the effect of methyl-isobutyl-ketone, MIBK, at various concentrations on glass transition temperature.18 Several solvents were used in this study18 to determine if the additivity rule can be useful to predict the glass transition temperature of a polymer-solvent system. The results, as a rule, depart from the linear relationship between glass transition temperature and solvent concentration. Generally, the better the solvent is for a particular polymer the higher is the departure. Hydrogen bond formation as a result of the interaction of polymer with solvent was found to contribute to changes in the electric properties of polyaniline.19 Hydrogen bonding causes changes in conformal structure of polymer chains. This increases the electrical conductivity of polyaniline. Water is especially effective in causing such changes but other hy-

704

George Wypych

drogen bonding solvents also affect conductivity.19 This phenomenon is applied 99 in antistatic compounds which require a 98 certain concentration of water to perform their function. 97 Surface properties such as smooth96 ness and gloss are affected by the rheological properties of coatings and solvents play 95 an essential role in these formulations. Sur94 face composition may also be affected by 2% DMS 1% DMS solvent. In some technological processes, 93 the surface composition is modified by 92 small additions of polydimethylsiloxane, 0 2 4 6 8 10 PDMS, or its copolymers. The most well Cyclohexanone content, v% known application of such technology has Figure 12.1.41. Surface DMS vs. cyclohexanone been the protection of external surface of concentration for two bulk contents of DMS in the space shuttle against degradation using polystyrene. [Data from Jiaxing Chen, J A Gardella, PDMS as the durable polymer. Other appliMacromolecules, 31, No.26, 9328-36 (1998).] cations use PDMS to lower the coefficient of friction . In both cases, it is important that PDMS forms a very high concentrations on the surface (possibly 100%). Because of its low surface energy, it has the intrinsic tendency to migrate to the surface consequently, surface concentrations as high as 90% can be obtained without special effort. It is more difficult to further increase this surface concentration. Figure 12.1.40 shows the effect of a chloroform/toluene mixture on the surface segregation of PDMS. When the poor solvent for PDMS (toluene) is mixed with chloroform, the decrease in the mobility of the PDMS segments makes the cohesive migration of the polystyrene, PS, segments more efficient and the concentration of PDMS on the surface increases as the concentration of toluene increases. Increase in PDMS concentration further improves its concentration on the surface to almost 100%. This shows the influence of polymer-solvent interaction on surface segregation and demonstrates a method of increasing the additive polymer concentration on the surface. Solvents with higher boiling points also increase surface concentration of PDMS because they extend the time of the segregation process.20 Figure 12.1.41 shows the effect of addition of cyclohexanone to chloroform.20 This exemplifies yet another phenomenon which can help to increase the surface concentration of PDMS. Cyclohexanone is more polar solvent than chloroform (Hansen parameter, δp, is 3.1 for chloroform and 6.3 for cyclohexanone). The solvation of cyclohexanone molecules will selectively occur around PDMS segments which helps in the segregation of PDMS and in the cohesive migration of PS. But the figure shows that PDMS concentration initially increases then decreases with further additions of cyclohexanone. The explanation for this behavior is in the amount of cyclohexanone required to better solvate PDMS segments. When cyclohexanone is in excess, it also increases the solvation of PS and the effect which produces an increased surface segregation is gradually lost. The surface smoothness of materials is important in solvent welding. It was determined that a smooth surface on both mating components substantially increases the strength of the weld.21 This finding may not be surprising since weld strength depends on close surSurface DMS, molar %

100


705

face contact but it is interesting to note that the submicrostructure may have an influence on phenomena normally considered to be influenced by features which have a larger dimensional scale. Non-blistering primer is another example of a modification which tailors surface properties.22 The combination of resins and solvents used in this invention allows solvent to escape through the surface of the cured primer before the material undergoes transitions that occur at high temperatures. Additional information on this subject can be found in Chapter 7. A review paper23 examines the nucleophilic properties of solvents. It is based on accumulated data derived from calorimetric measurements, equilibrium constants, Gibbs free energy, nuclear magnetic resonance, and vibrational and electronic spectra. Parameters characterizing Lewis-donor properties are critically evaluated and tabulated for a large number of solvents. The explanation of the physical meaning of polarity and discussion of solvatochromic dyes as the empirical indicators of solvent polarity are discussed (see more on this subject in Chapter 10).23

REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

J F Rabek, T A Skowronski, B Ranby, Polymer, 21, 226 (1980). H Kubota, M Kimura, Polym. Deg. Stab., 38, 1 (1992). G Teissedre, J F Pilichowski, J Lacoste, Polym. Degrad. Stability, 45, No.1, 145-53 (1994). M Shah, N S Allen, M Edge, S Navaratnam, F Catalina, J. Appl. Polym. Sci., 62, No.2, 319-40 (1996). P Hrdlovic, I Lukac, Polym. Degrad. Stability, 43, No.2, 195-201 (1994). M L Jackson, B J Love, S R Hebner, J. Mater. Sci. Materials in Electronics, 10, No.1, 71-9 (1999). Hu Shengkui, A Mejiritski, D C Neckers, Chem. Mater., 9, No.12, 3171-5 (1997). H Ohkita, A Tsuchida, M Yamamoto, J A Moore, D R Gamble, Macromol. Chem. Phys., 197, No.8, 2493-9 (1996). G Wypych, Handbook of Material Weathering, ChemTec Publishing, Toronto, 1995. R D Scurlock, D O Martire, P R Ogilby, V L Taylor, R L Clough, Macromolecules, 27, No.17, 4787-94 (1994). A T Soltermann, D de la Pena, S Nonell, F Amat-Guerri, N A Garcia, Polym. Degrad. Stability, 49, No.3, 371-8 (1995). K P Ghiggino, J. Macromol. Sci. A, 33, No.10, 1541-53 (1996). M Swistek, G Nguyen, D Nicole, J. Appl. Polym. Sci., 60, No.10, 1637-44 (1996). L Guy, B Fixari, Polymer, 40, No.10, 2845-57 (1999). J Jeczalik, J. Polym. Sci.: Polym. Chem. Ed., 34, No.6, 1083-5 (1996). W K Czerwinski, Macromolecules, 28, No.16, 5411-8 (1995). N Baute, C Calberg, P Dubois, C Jerome, R Jerome, L Martinot, M Mertens, P Teyssie, Macromol. Symp., 134, 157-66 (1998). S Turri, M Scicchitano, G Gianotti, C Tonelli, Eur. Polym. J., 31, No.12, 1227-33 (1995). E S Matveeva, Synthetic Metals, 79, No.2, 127-39 (1996). Jiaxing Chen, J A Gardella, Macromolecules, 31, No.26, 9328-36 (1998). F Beaume, N Brown, J. Adhesion, 47, No.4, 217-30 (1994). M T Keck, R J Lewarchik, J C Allman, US Patent 5,688,598, Morton International, Inc., 1996. Ch Reichardt, Chimia, 45(10), 322-4 (1991).

706

Ranieri Urbani and Attilio Cesàro

12.2 CHAIN CONFORMATIONS OF POLYSACCHARIDES IN DIFFERENT SOLVENTS Ranieri Urbani and Attilio Cesàro Department of Biochemistry, Biophysics and Macromolecular Chemistry, University of Trieste, Italy

12.2.1 INTRODUCTION Carbohydrate monomers and polymers are present in all living organisms and are widely used in industrial applications. In life forms they show and express very diverse biological functions: as structural, storage and energy materials, as specific molecules in the immunochemistry of blood, as important polymers of cell walls determining cell-cell recognition,1 antigenicity and viral infection, etc.. The variability of primary structure and conformation makes the carbohydrate molecule extremely versatile, for example, for specific recognition signals on the cell surface.2 In any biological system whatsoever the shape and size adopted by carbohydrates in different solvent environments have been widely demonstrated to be responsible for the biological function of these molecules. In recent years, significant progress has been made in the improvement of both the experimental and theoretical research tools needed to study the conformational complexity of carbohydrates in solution, such as X-ray and neutron scattering techniques (SAXS and SANS), atomic force microscopy (AFM), high-resolution NMR spectroscopy and relaxation techniques, and computational methods. All the experimental and computational methods unequivocally indicate the relevance of the environment (e.g., solvent composition, pH and salt conditions, temperature) on the topological shape and the properties of the carbohydrate solutes. The general problem of solvent effect on the conformational states and the preferential solvation of oligo- and polysaccharides has been tackled mainly to validate detailed molecular models which were developed for relating the structural characteristics of these macromolecules to their chemical, physical, and biological properties in solution.3-5 Semi-empirical methods (unrefined in the sense that molecular parameters are not adjusted for the specific case studied) have been proven useful and generally applicable to different chain linkages and monomer composition. However, it must be clear that all these methods, unless specifically stated otherwise, refer to calculations of the unperturbed chain dimension and therefore do not take into account the excluded volume effects (which arise typically from long-range interactions). In most cases, solvation effects (which are short range) are not explicitly taken into consideration, e.g., molecular parameters of the solvent do not enter in the calculations, although some exceptions are found in literature. The rationale for the correct setting of current knowledge about the shape of polysaccharides in solution is based on three factors: the correlation between primary structure (i.e., the chemical identity of the carbohydrates polymerized in the chain), intrinsic conformational features dictated by the rotational equilibria (often the major contributions are due to the rotation about the glycosidic linkages) and the interaction with the other mo-

12.2 Chain conformations of polysaccharides

707

lecular species in the system (mainly the solvent which determines, therefore, the solubility of the chains). In this chapter the description of the solvent effect is given within the framework of some specific experimental results and computational methods for studying and predicting oligo- and polysaccharide conformations in solution. It is not the authors’ intention to make an in depth investigation into the general methodologies which have been widely reported over recent years (and in this book) but rather to provide a step-wise presentation of some conformational features which have upheld theoretical predictions with experimental observations. The number of examples and approaches is necessarily limited and the choice undoubtedly reflects the authors’ preferences. Nonetheless, the aim is to be as informative as possible about the conceptual difficulties and conceivable results. 12.2.2 STRUCTURE AND CONFORMATION OF POLYSACCHARIDES IN SOLUTION 12.2.2.1 Chemical structure The primary structure of polysaccharides (glycans) is often complicated by different kinds of linkages in homopolymers and different kinds of monomeric units, which give rise to a huge number of different polymers. Glucans (see a general formula in Figure 12.2.1) are those composed exclusively of glucose, while glucuronans are polymers of glucuronic acid. Similarly mannans and galactans as well as mannuronans and galacturonans, are homopolymers of mannose, galactose, mannuronic acid and galacturonic acid, respectively. Although all the polysaccharides discussed here show a structural regularity, they may not be simple homopolymers. Their chemical structure can sometimes be fairly complicated.

Figure 12.2.1. Example of the chemical structure of a polysaccharidic chain as a sequence of β-(1-4)- linked D-glucose units having a side chain of β-glucose linked (1-6) to the backbone. The glycosidic dihedral angles are also indicated.

12.2.2.2 Solution chain conformation A regularity of primary structure could imply that the chains may assume ordered helical conformations, either of single or multiple strand type, both in the solid state and in solution. A knowledge of both polysaccharidic chain structure, up to the three-dimensional molecular shape, and the interaction of the polymers with other molecular components, is essential, in order to understand their capability to form supramolecular structures, including physical gels, with specific rheological properties, which have important implications for controlling and upgrading properties in industrial applications. The rationale is that the physico-chemi-

708


Figure 12.2.2. (a) Snapshot of random chains of homoglucans obtained from Monte Carlo calculation and (b) ordered helical structures of the same polymers as measured by X-ray diffraction method.

cal properties, which form the basis of the industrial applications of polysaccharides are directly related to the structure/conformation of the chain. Many biological macromolecules in the solid state assume regular helices, which can be represented by means of a few geometrical parameters and symmetry relations. These helical structures originate from the stereo regularity of backbone monomers and are easily described, conceptually simple, and therefore used (all too often) as an idealized model for all the actual shapes. Elements of helical regularity are essential in the description of the structure of nucleic acids and polypeptides. Some biopolymers, e.g., globular proteins, almost completely preserve their structural regularity in solution. Such globular structures are not known in polysaccharides. Nonetheless, helical conformations have been proposed to represent the structure of many polysaccharides, microbial glycans in particular. An excellent overview of the stereo-regular helical conformations of polysaccharides deduced from X-ray fiber diffraction studies has recently been published.6


709

However, thermodynamic arguments suggest that a partially disordered state is an essential prerequisite for the stability of polymeric systems in solution. Under these circumstances, realistic chain pictures of polysaccharides will not necessarily be generated from the repetition of a single conformational state, which is usually identified by the minimum energy state found in the internal conformational energy calculations. Statistical approaches to these conformational energy surfaces7,8 suggest a more disordered solution conformation (Figure 12.2.2, snapshots marked with a) than the chain structures proposed so as to fit the helical regularity deduced from x-ray fiber diffraction studies (Figure 12.2.2, structures marked with b). Thermal fluctuations are in general sufficient to generate delocalized disorder, unless diffuse interactions, cooperative in nature, ensure long-range order. From the experimental point of view, several polysaccharides with different chain linkage and anomeric configuration have been studied to determine to what extent the polymeric linkage structure and the nature of the monomeric unit are responsible for the preferred solvation and for the chain topology and dimensions.9 Conversely, since it is generally understood that the structure and topology of many macromolecules are affected by solvation, theoretical models must include these solvent effects in addition to the internal flexibility, in order to estimate changes in the accessible conformations as a result of the presence of the solvent molecules. The concept of chain conformational disordering and dynamics in solution is associated with the existence of a multiplicity of different conformations with accessible energy and, moreover, exhibiting their topological differentiation. The above conformational variability of polymeric chains is implicitly recognizable by the great difficulty in crystallization and by the typical phenomenon of polymorphism. This discussion is relevant in particular to ionic polysaccharides (see below) because, from the polyelectrolytic point of view, the transition from a more compact conformation (also an “ordered chain”) to an extended coil conformation, is usually associated with a net variation in ionic charge density along the chain. Polysaccharides generally dissolve only in strongly solvating media. Water displays a complicated behavior: it is a good solvent for monomers and oligosaccharides inasmuch as it is able to compete with the specific inter- and intra-molecular hydrogen bond network (Figure 12.2.3). In many cases it is the thermodynamic stability of the solid-state form which protects the solute molecules from being solubilized.10 Nevertheless, some other strong solvents, such as, dimethylsulfoxide, DMSO, and 1,4-dioxane are known to be good solvents for carbohydrate polymers. Many commercial applications of polysaccharides require compatibility with different solvents and solutes (organic solvents, salts, emulsifiers, plasticizers, enzymes, etc.), for example, in pharmaceutical matrices, paints and foods. In this field, solvent compatibility of some glycans has been improved and controlled by functionalization and derivatization in order to obtain a proper degree of substitution, which determines a wide range of compatibility properties. At the molecular level, various specific and non-specific solvent-solute interactions may occur in polysaccharide solutions that may result in a change in the conformational shape, solubility, viscosity and other hydrodynamic and thermodynamic properties. Hydrophilic interactions such as hydrogen bonding and electrostatic interactions are believed to be factors that influence the conformation of polysaccharides in solution, although the question is being raised (more and more) as to the implication of patches of hydrophobic intermolecular interactions, especially for chain aggregations. One important feature is the

710


Figure 12.2.3. Role of structural, conformational and substituent features on the solution properties of oligo- and polysaccharides.

surface that saccharide segments address to solvent molecules that allows a high degree of favorable interactions. The water structuring in the solvation of polysaccharides also contributes to the stability of saccharides in solution,11 which may be altered by competition of other co-solutes or co-solvents which are able to modify the extent of hydrogen-bonded interactions among components. Urea, for example, is considered a breaker of ordered polysaccharide conformation in gel networks, such as those obtained with agar and carrageenan as well as with microbial polysaccharides. Over the last twenty years or so, several approaches have been made to determine the three-dimensional structure of oligosaccharides and theoretical calculations are becoming


711

an increasingly important tool for understanding the structure and solution behavior of saccharides.11-15 Although computer facilities and calculation speeds have grown exponentially over the years some of these techniques, like quantum mechanic methods (ab initio methods), have been shown not to be useful in dealing with the complexity of systems which involve many atoms such as in macromolecular systems. On the other hand, these techniques have been successfully applied to small molecules, e.g., mono- and disaccharides, in predicting charge distribution on the atoms, conformations and transitions among accessible conformations,16 thus providing a background knowledge for the more complicated systems. 12.2.3 EXPERIMENTAL EVIDENCE OF SOLVENT EFFECT ON OLIGOSACCHARIDE CONFORMATIONAL EQUILIBRIA The problem of sugar conformation and dynamics in solution is related to the question of to what extent are oligo- and polysaccharides intrinsically flexible under the different experimental conditions. The answer to this question, which requires a complete knowledge of the time-space dependence of the chain topology,17 is often “rounded-off” by the use of empirical terms like “flexibility”. A further problem is to what extent does the solvent contribute to stabilizing some conformational states rather than others. Solution properties are functions of the distribution of conformations of the molecules in the solvated states, in the sense that the experimental data are statistical thermodynamic averages of the properties over all the accessible conformational states of the molecule, taking each state with a proper statistical weight. This aspect can be better illustrated by taking into consideration the accessible conformational states of a simple sugar unit and the conformational perturbation arising from the changes in the interactions with the surrounding solvent medium. For an α-pyranose ring, for example, the rotation about carbon-carbon bonds and the fluctuations of all ring torsional angles give rise to a great number of possible conformers with different energies (or probabilities). Some of these are identified as the preferred rotational isomeric states in various environments. The boat and boat-skew conformers of a pyranose ring, which are higher in energy than the preferred chair form, correspond to a major departure from the lowest energy chair conformation as illustrated by the globular conformational surface of Figure 12.2.4. Additional conformational mobility in a monosaccharide is due to the rotations of exocyclic groups, namely, OH and especially CH2OH. Let us point out that in a polymer these ring deformations do not normally occur, but, when they do, they may determine to a great extent the equilibrium mean properties and the overall mean chain dimensions.18,19 However, one very recent theory is that the elastic properties of amorphous polysaccharides are related to the glycosidic ring deformation.20 D-ribose is probably the best example of sugar that reaches a complex conformational equilibrium, giving rise to the mixture composition shown in Figure 12.2.5. The percentage of each form is taken from Angyal’s data21,22 with integration of the sub-splitting between the 4C1 and the 1C4 forms. The prime purpose of this analysis is to point out that the stability of each conformer is not determined solely by the intrinsic internal energy, which can be evaluated by means of e.g., ab-initio quantum mechanics calculations, but is strongly influenced by all the solvation contributions. Therefore, the conformer population may be shifted by changing temperature, solvent composition, or any other external variable such as, for example, adding divalent cations (see Table 12.2.1). The evaluation of the actual concentration of the several conformers involved in the equilibrium immediately leads to

712


Figure 12.2.4. Representation of all possible conformations of the pyranosidic ring and respective inter-conversion paths.

Figure 12.2.5. D-ribose conformational equilibria.


713

the partition function of the system (i.e., the free-energy change or the equilibrium constant for a transition between conformers). Table 12.2.1. Percentage composition of furanose and pyranose forms of D-ribose and glucose in different solvent systems Sugar

Solvent H2O, 31oC o

D-ribose

D2O, 30 C o

DMSO, 30 C 2+

Ca , 1.27 M

α-F

β-F

α-P

(4C1) β-P

(1C4) β-P

13.5

6.5

21.5

44.0

14.5

8.0

14.0

23.0

41.0

14.0

6.0

22.0

16.2

31.0

24.8

13.0

5.0

40.0

14.0

28.0

45.0

55.0

45.0

53.0

o

Glucose

DMSO, 17 C o

pyridine, 25 C

0.6

1.0

Passing from monomer units to oligomers (disaccharides and higher oligosaccharides), the dominant features of molecular flexibility become those due to rotations about the glycosidic linkages. Although other conformational fluctuations may contribute to the local dynamics of the atoms or group of atoms, only the glycosidic linkage rotations are able to dramatically change the conformational topology of oligomers at ambient temperature. The aim of a conformational analysis of oligosaccharides is thus to evaluate the probability (that is the energy) of all mutual orientations of the two monosaccharidic units, as a function of rotations about the glycosidic linkages, defined by the dihedral angles: ϕ = [H1-C1-O1-Cm] ψ = [C1-O1-Cm-Hm] where m is the aglycon carbon number in the reducing ring. The important region of ϕ and ψ rotations is the one with energy variations in the order of kT, the thermal motion energy, because this may produce a large ensemble of accessible conformational states for the oligosaccharide. Even when the rotational motion is restricted to only a few angles, the fluctuations of many such glycosidic bonds is amplified along the chain backbone, as the molecular weight increases. The accumulation of even limited local rotations may produce very large topological variations in the case of polymeric chains and consequently relevant changes in thermodynamic, hydrodynamic and rheological properties of these systems. Other internal motions often make only small contributions to the observable properties on the macromolecular scale.23 Experimentally, NMR techniques are among the most valuable tools for studying conformations and dynamics of oligosaccharides in solution by determining chemical shifts, coupling constants, NOE’s and relaxation time.24-30 In the study of saccharide conformations the potential of coupling constants evaluation, specially the (hetero-nuclear) carbon-proton spin-spin coupling constants, nJC,H, and their dependence on solvent medium is well known. Several empirical correlations between 1JC,H and structural parameters like dihedral angles have been reported,31 although the low natural abundance of the 13C isotope often made the measurements technically difficult and 13C enrichment was required for the generation of data of sufficient quality for the quantitative analysis. Nowadays, due to the progress of FT-NMR spectrometers, measurements of the three-bond (vicinal) carbon-pro-

714


ton coupling constant, 3JC,H have indeed been made possible on the natural-abundance 13C spectra. Much endeavor has been invested in establishing a general Karplus-like relation for the angular dependence of nJC,H, especially on the dihedral angles in the glycosidic region because the overall three-dimensional structure of oligo- and polysaccharides are related to the glycosidic features in terms of dihedral (rotational) angles. From comparison of experimental data and theoretical results (using quantum-mechanical and semi-empirical methods) on model compounds,32-34 the angular dependence of the coupling constants, nJC,H, on the anomeric and aglycon torsional angles ϕ and ψ,31,34 and on the dielectric constant of the solvent35 can be written in a general form such as: 1

JC , H = A1 cos 2χ + A2 cos χ + A3 sin 2χ + A4 sin χ + A5 + A6 ε

3

JC , H = B1 cos 2 χ + B 2 cos χ + B 3 sin χ + B 4 sin 2χ + B 5

[12.2.1]

where: Ai, Bi χ ε

constants different for α and β anomers ϕ or ψ solvent dielectric constant

Experimental nJC,H values for conformationally rigid carbohydrate derivatives allow to calculate the constant values of Ai and Bi in equations [12.2.1]. The major practical use of these equations is their ability to estimate the glycosidic dihedral angles from experimental n JC,H data, in combination with other complementary results, for example, from NOESY,36 X-ray and chiro-optical experiments. Since the experimental values are averaged over all the accessible conformational states in solution, they do not necessary reflect the property of only the most probable conformer23 but they nevertheless include contributions of all the conformers, each one taken with its proper statistical weight. Thus, the quantitative interpretation of experimental data in terms of accessible conformational states of flexible molecules requires the additional theoretical evaluation of the energy of the molecule as a function of internal coordinates. Since the dependence of the observed coupling constants on the conformation is non-linear, it derives that: n

JC , H (χ) ≠ n JC , H ( χ

)

Only for a linear dependence of nJC,H on χ, the equation could hold the equals sign. Because of the simultaneous dependence of the property on χ and on the complexity of potential energy function, E(χ), the ensemble average of , as well as of any property of interest, can be calculated only by taking into consideration the conformational energy surface. Apart from NMR and the methods suitable for characterization of overall chain dimensions (which we will touch on below), there are not many simple experimental techniques that can be used to study sugar conformation and that are directly correlated to theoretical results based on calculated potential energy surfaces or force fields. An exception is given by chiro-optical techniques, which provide important (although empirical) structural information as optical rotation experiments have been shown to be very useful and informative in the study of saccharides in solution.37,38 Literature gives the experimental evidence for the effects of the external conditions (solvent and temperature) on the optical rotation. The additivity methods, proposed in the Fifties by Whiffen39 and Brewster,40 were


715

extended by Rees and co-workers,37,41 who derived expressions for the contribution to the optical activity of changes in conformational states of oligo- and polysaccharide chains. More recently, Stevens and co-workers42,43 developed a computational model for optical activity and circular dichroism based on the Kirkwood theory in which the calculations of the lowest energy component of a molecular σ − σ * transition, which derives from the mutual interaction of all σ − σ * transitions on C-C, C-H and C-O bonds, are carried out. The calculated optical properties are therefore geometry-dependent and the optical properties have been calculated as statistically averaged properties over an ensemble of all possible conformations, theoretically obtained as a function of ϕ and ψ. This model also provides a useful tool for testing the quality of force field parametrizations and the possibility of some refinement of force constants for a non-explicit inclusion of solvent environment. 12.2.4 THEORETICAL EVALUATION OF SOLVENT EFFECT ON CONFORMATIONAL EQUILIBRIA OF SUGARS 12.2.4.1 Classical molecular mechanics methods Theoretical approaches to the conformational analysis of oligosaccharides in solution become inherently more complicated than those for monomeric sugars, at least for the computer time required to optimize the structures. Depending on the degrees of freedom taken into consideration and on the level of sophistication of the method used, the conformational analysis may give different results. It has been the custom in the past to compute the conformational energy as a function of the dihedral angles ϕ and ψ only, by assuming the sugar ring to be rigid (rigid-residue method). More recently, all the internal coordinates are allowed to adjust at each increment of ϕ and ψ, relaxing the structure toward a local minimum (relaxed-residue method). The rigid-residue approach is still considered suitable as a starting point, although some warning must be alerted to the use of the same set of structural coordinates for all sugars, as was initially done. The approach may however be useful as a starting point in the conformational analysis of polysaccharides, provided that the coordinates of all the atoms in the monomeric unit have been calculated by a suitable independent method.13, 44 Figure 12.2.6 reports, as a general example, the rigid-residue energy surfaces of two representative disaccharides, namely α-(1-4)-D-glucopyranosil-D-glucopyranose (D-maltose, Figure 12.2.6a) and the β-(1-4)-D-glucopyranosil-D-glucopyranose (D-cellobiose, Figure 12.26b). Both maps present multiple minima separated by barriers that are only a few kilocalories high. For many disaccharides the barriers between minima can be very high and they are often overestimated in the rigid-residue approximation. In general, the conformational energy map shows that only a limited portion of the total ϕ, ψ conformation space is actually accessible to the dissaccharide at room temperature. The dissaccharide is not frozen into its lowest energy conformation, nevertheless the steep walls of the allowed region of conformational energy may dramatically limit the multiplicity of thermally accessible conformational states. One useful method, for taking into account the effect of solvent media upon the conformational properties of glycosidic structures, is the continuum reaction field method.45, 46 This method is based on the Scaled Particle Theory (SPT) equations47 and on Onsager’s theory of the reaction field, as applied by Abraham48 by considering the solvent as a dielectric continuum. In this approach the total conformational energy, Gtot, is given by: G tot = G conf + Gsolv

716


Figure 12.2.6. Conformational energy maps calculated for (a) D-maltose and (b) D-cellobiose. Contours are in kcal·mol-1 with increments above the minimum (x) of 0.5 kcal·mol-1.

where the total conformational free energy, Gtot, is given by the sum of the contribution due to the in vacuo conformational state energy, Gconf, and the solvation contribution, Gsolv. The latter term is a sum of contributions due to the energy required to create, in a given solvent, a cavity of suitable size to accommodate the solute molecule in a given conformational state and the interaction energy between the solute and the surrounding solvent molecules, as schematically shown in Figure 12.2.7:49 Gsolv = G cav + G el + G disp [12.2.2] The expression for Gcav, that is the free energy required for the formation of a cavity (first step in Figure 12.2.7), at a temperature T and a pressure P, is taken from the Scaled Particle Theory (SPT) which has been successfully applied in the study of thermodynamic properties of aqueous and non-aqueous solutions:47,50

Figure 12.2.7. General scheme of the theoretical approach for the solvation energy calculation.


717

2  3y  $  3y G cav yP $ 3 9 y   $ 2   R +  R +  +  = − ln (1 − y) +  R 1− y 2  1− y   RT ρkT  1− y   

[12.2.3]

where: y av, au $ R ρ R k

= 4 πρa 3v / 3, the reduced number density radii of hard sphere equivalent solute (u) and solvent (v) molecules = au / av solvent density gas constant Boltzmann constant

It is noteworthy that Gcav is a function solely of solvent density and solvent and solute dimensions and represents a measure of the cohesive forces among solvent molecules. In a given solvent, this contribution is a function of the radius of solute molecule which may differ remarkably as a function of dihedral angles ϕ and ψ only. The electrostatic interaction term Gel between solute and solvent is based on the continuum reaction field48 which takes into account a reaction potential induced by the solute dipole and quadrupole (third step in Figure 12.2.7) in a continuum medium of dielectric constant ε: G el =

 3HX KX  bF  + bF 1 − exp  −  1 − lX 5 − X  16RT  

[12.2.4]

with: K=

µ 2u a

3

, H=

Q u2 a

9

, l=

(

n +2 2 u

),

X=

ε −1 2ε + 1

ε≤2

0 for

F=

5

2 n u2 − 1

(ε − 2)(ε + 1)    ε   1/ 2

 T  b = 4.35   300 

(

1/ 2

a 3 / 2 ruv3

for

)

3

ε>2

 K + Ha 2   r2  uv

1/ 2

   

where: µu Qu nu a ruv

solute dipole moment solute quadrupole moment solute refractive index = av +au = a/21/2

ruv is the average distance between the solvent and the solute molecule defined in terms of the radius of cavity a.

718


The free energy of dispersion, Gdisp, in equation [12.2.2] takes into account both attractive and repulsive non-bonding interactions and is expressed as a combination of the London dispersion equation and Born-type repulsion:51 G disp = −0.327Nv α u α v

I u Iv −6 ruv I u + Iv

[12.2.5]

where: α I Nv

molecular polarizability ionization potential nearest-neighbor solvent molecules

Nv is the number of molecules surrounding the solute molecule in a given conformation and is calculated from the following equation:52

(

N =  a + 2a v  u v

)

3

 4πNA − a    3V  v 3 u

  

3

where: Vv NA

solvent molar volume Avogadro number

In order to elucidate the effect of different solvents on the conformation, the energy differences between conformers in a given solvent are more relevant than the absolute solvation energies in each solvent. In particular, it is important that the perturbation effect on the detailed shape in the low energy regions of the conformational map. Figure 12.2.8 shows the free-energy of solvation and energy contributions for the maltose dimer as a function of ψ calculated at ϕ=-30° and refers to the energy of the (-30°, 180°) conformer for two solvents, water and DMSO (Figure 12.2.8 a and b, respectively). In this section, the cavity term is a complex function of the size of the maltose molecule, as determined by the spatial orientation of the two glucose residues. The cavity energy is, by definition, always unfavorable; the more expanded the conformers, which are usually located in the low energy region of the map (around ψ=0° for maltose), the larger the Gcav due to the size of larger cavities to be created in the solvent. The electrostatic free-energy, Gel, increases with the dipole and quadrupole moments of solute molecule and decreases with the radius of the cavity. On the other hand, the dipole moment µ is a function of dihedrals ϕ and ψ and for maltose has two large maxima around (ϕ, ψ)=(-40°,0°) and (180°, 140°) which is in agreement with the major contributions of Gel shown in Figure 12.2.8. The electrostatic free-energy (eq. [12.2.4]) is also dependent on the solvent dielectric constant and increases in passing from DMSO (ε=46.68 debye) to water (ε=78.30 debye) at 25°C. The dispersion term (eq. [12.2.5]) makes a significant contribution (-20 to -70 kJ·mol-1) to the absolute value of solvation free-energy but the angular dependence is very small giving an almost equal contribution to the energy of conformers. In an early paper53 it was shown that, in the comparative cases of cellobiose and maltose, the probability distribution of conformers (Figure 12.2.9) is affected by the presence of the solvent, changing the shape of the function from one solvent to another. The minimum of the maltose map goes from (-20°,-30°) to (-10°,-20°) in water and DMSO, while that of cellobiose map goes from (0°, 50°) to (-30°, -20°) in both solvents.53 These solvent perturbations (apparently small) on the conformational energies have a great effect on the proba-


719

Figure 12.2.8. Section of the maltose energy map at ϕ=-30° showing the dependence of contributions to the Gsolv as a function of angle ψ in water (a) and DMSO (b).

Figure 12.2.9. Histograms of probabilities of conformational states for cellobiose in vacuo (a), water (b) and DMSO (c), and for maltose (a’, b’, and c’).

720


bilities associated with the conformers and therefore produce a rearrangement of statistical weights among conformers and, thus, a variation of the average properties of the system, as for example illustrated by the changes of the chain dimension parameters (see below Figure 12.2.11). Calculated average data like dipole moment, µ, linkage rotation, Λ, and proton-carbon coupling constants across glycosidic linkage, 3JC1-H4 and 3JC4’-H1, for β-D-maltose14 and β-D-mannobiose45 in different solvents are reported in Table 12.2.2. These results show that the equilibrium composition of dimer conformers depends strongly on the solvent and that the departure from the in vacuo conformation increases with increasing solvent dielectric constant. For mannobiose the determining factor of the solvent effect on the conformation is the intra-residue electrostatic interaction, which depends on ϕ and ψ in the same manner as the dipole moment. Table 12.2.2. Calculated average values of dipole moment, four-bonds proton-carbon coupling constant and linkage rotation of β-D-maltose and β-D-mannobiose in different solvents at 25°C Solvent

ε

ϕ, deg

ψ, deg

µ, debye

3

JC1-H4, Hz

JC4'-H1, Hz

Λ, deg

3

solute: β-D-maltose vacuum

-

-21

-28

3.80

4.2

4.5

-31

1,4-dioxane

2.21

-22

-22

3.89

4.1

4.5

-30

pyridine

12.40

-24

-39

4.02

4.0

4.5

-27

ethanol

24.55

-24

-41

4.09

3.9

4.5

-27

methanol

32.70

-25

-46

4.17

3.9

4.5

-25

DMSO

46.68

-24

-41

4.10

3.9

4.5

-28

water

78.30

-28

-65

4.47

3.7

4.7

-19

solute: β-D-mannobiose vacuum

-

81

-18

6.00

2.29

4.14

77

1,4-dioxane

2.21

82

-19

5.92

2.23

4.15

76

pyridine

12.40

91

-22

6.01

1.95

4.22

69

methanol

32.70

97

-24

6.08

1.72

4.25

54

DMSO

46.68

94

23

6.05

1.84

4.23

67

water

78.30

100

-18

6.27

1.53

4.32

57

It should be pointed out that even the small variations in the dihedral angles due to the presence of solvent molecules may produce a large change on a macromolecular scale, as discussed above. 12.2.4.2 Molecular dynamic methods One of the most powerful theoretical tools for modeling carbohydrate solution systems on a microscopic scale and evaluating the degree of flexibility of these molecules is the molecular dynamics technique (MD) which has become popular over the last two decades. The first reported works of MD carbohydrate simulation appeared in 198654,55 and since then an in-


721

creasing number of MD simulations on sugars has been carried out.11,13,56,57 The explicit representation of solvent molecules is required especially in biological systems, where solvent structuring plays an important role. Starting with an appropriate potential energy function for sugar-water interactions,13,58 the common solution simulations are carried out by placing the solute molecule in the center of a cubic box of finite dimensions containing a given number of solute molecules which usually corresponds to at least three solvation shells.11 The macroscopic system is then simulated by using the approximation known as “periodic boundary conditions”,59 in which the entire box is replicated in every direction, leaving solute and solvent molecules to interact with each other both in central and in replica boxes and setting the long-range interactions to smoothly decrease to zero by using the appropriate switching functions. A pair distribution function, g(r), defined as: g(r ) =

1 dN(r ) 4πρr 2 dr

where: r ρ N(r)

interatomic distance the bulk number density number of atoms of given type at distance r

has been used to evaluate the normalized probability of finding a water oxygen atom at a distance r from a given atom on the carbohydrate molecule.11 In this way the anisotropic distribution of solvent molecules around carbohydrate solutes was reported11,60,61 showing an exceptional structuring of water molecules which extends to greater distances around the solute as compared to the pure solvent. One of the most interesting results of these simulations is the identification of the spatial distribution of water molecules on the van der Waals’ surface of the carbohydrate molecules. Figure 12.2.10 shows a probability density excess of the water molecules in a channel, which is effectively a bisector of the two closest sites for hydrogen bonding. When the energy maps have been obtained by molecular mechanics calculations, dynamics simulations of disaccharides in various conformations are carried out to analyze the

Figure 12.2.10. Contours of solvent anisotropic density around the α , α-trehalose disaccharide obtained from MD simulation [unpublished figure kindly provided by J.W. Brady and Q. Liu].

722


typical motions of the molecules along the conformational space at moderate temperatures. In general, initial conformations in the simulations are taken from one of those of minimized geometry as identified in the calculation of the energy map. Trajectories are then computed by assigning to the atoms velocity components randomly selected from a thermal distribution at a given temperature. By superimposing the trajectories of fluctuations of ( ϕ, ψ) on the energy map, it is possible to observe a variety of motions, involving both the structures of the molecular rings and in some cases rotations about the glycosidic bonds and exocyclic torsions.62 In the case of multiple minima, conformational transitions between low energy regions may also be revealed. 12.2.5 SOLVENT EFFECT ON CHAIN DIMENSIONS AND CONFORMATIONS OF POLYSACCHARIDES In addition to the change of entropy, minimization of free energy in a binary dilute solution containing solvent and polymer occurs as a result of the favorable interactions between the chain segments and the solvent, which replace the homotactic interactions between solvent molecules and those between chain segments. In a good solvent and a very dilute solution, it is likely that only solute-solvent and solvent-solvent interactions prevail. However, in a bad solvent (and, in general, in a more concentrated solution) persistence of segmental interactions among chains is the major contribution in the macroscopic properties of the system. The theoretical evaluation of the entropic contributions arising from the configurational nature of the chain molecule is possible on the basis of thermodynamic-statistical models.63 More troublesome is the contribution of enthalpy change of mixing, ∆ mix H, of non-ionic polysaccharides and water which cannot be predicted, not even in sign. Furthermore, the scarcity of literature data does not allow any empirical rationalization although, in most cases, contributions significantly smaller than the related monomers are reported. Notwithstanding the above limitations, a general picture can be drawn showing that the average dimension of a chain, experimentally obtained with light scattering or viscometric measurements, depends upon solvent interactions and behavior, in addition to the intrinsic features of the polysaccharide (chemical nature and linkage of monomers, conformational equilibria, etc.). The basic axiom is that solution properties are strictly related to the conformation of the molecules in the solvated state and this state, in turn, is only statistically defined from the primary (chemical) structure. In first instance, one can generalize the statement that non-ionic crystalline molecules and especially polymers, barely preserve their ordered conformation upon dissolution (Figure 12.2.2). Polymers and oligomers (including most carbohydrate molecules) generally assume a statistically disordered conformation in solution since, in the absence of specific favorable enthalpy contributions, polymer-solvent interactions provide a small increment in the entropy of mixing. Under these very common circumstances, the dissolution of a crystalline carbohydrate molecule, which is stabilized in the solid state by a great number of interactions, becomes a thermodynamically unfavorable event. Structurally speaking, the dissolution is made possible only by a sufficient increment in conformational entropy, which is described by the increment in conformational states accessible to the molecule. In general, some conformational features, resembling those observed in the solid state, can be preserved also in solution (Figure 12.2.2), but the degree of order is strictly related to the presence of the solvent, in addition to the temperature and to the entropy of mixing. Figure 12.2.2 shows glucan chains with different types of linkages, and different pictorial trajectories which give different and sometimes surprising values of configurational entropy


723

are also observed. For example, those of (1-4)-linked α-D-glucan and of (1-3)-linked β-D-glucan seem fairly restricted to some pseudo-helical character compared to the more disordered set of trajectories that would be obtained if rotations about the glycosidic bonds were completely unrestricted. Possible interactions between residues of the polysaccharide chain that are not nearest-neighbors in the primary sequence of the polymer can sometimes be ignored. In this case, a computer-based polysaccharide chain can be constructed from the conformational energy map of the dimeric units. The Monte Carlo method64 and the Flory matrix methods65 are commonly used in the so-called “nearest-neighbors approximation” to mimic the polymer chains in the pure amorphous state or in dilute solution. The Monte Carlo sample of chains reflects the range of conformations experienced by any single chain as a function of time or, equivalently, the range of conformations in a large sample of chemically identical polymer molecules at any instant in time. In either sense, the sample can be analyzed to deduce both the characteristics of individual chain conformations and the mean properties of the sample as a whole, which correspond to those in the equilibrium state of the chain. Results refer, however, to an “unperturbed” chain model that ignores the consequences of the long range excluded volume effect, because only nearest-neighbor interactions are accounted for. Given a sufficient Monte Carlo sample of chains in equally probable representative conformations, it is possible to assess many mean properties of the polymer in question simply by computing numerical (unweighted arithmetic) averages over the number of chains in the sample. For example, the mean square end-to-end distance, the mean square radius of gyration, or the angular dependence of scattered light (particle scattering factor) are all mean geometric properties readily computed from a knowledge of the coordinates of the atoms or atomic groups which are generated in the Monte Carlo sample. The average topological properties are described through the chain-length dependence of quantities such as the characteristic ratio, Cn, the persistence length, Pn, and the correlation function, Fn, defined as: r  L1  n r Cn = , Pn =   ∑ L i  L1  i =1 nL2   r2

rr , Fn = uu n

0

0

[12.2.6]

0

where: n L 0

number of saccharide units average virtual bond length mean square end-to-end distance

The virtual bond vector is often defined for each monosaccharidic unit as connecting oxygen atoms involved in glycosidic linkages. Figure 12.2.11 shows the calculated properties of equations [12.2.6] for two homoglucan chains which differ only in the configuration of the anomeric carbon, i.e., (a) the [α-D-(1-3)-glc]n (pseudonigeran) and (b) the [β-D-(1-3)-glc]n (curdlan). Whatever the chemical features, provided that the molecular weight is very large (that is for a degree of polymerization n which approaches infinity), the distribution of unperturbed polymer end-to-end length is Gaussian and Cn equals an asymptotic characteristic ratio C∞, i.e.:

724


C∞ = lim

n→∞

r2 nL2

=

6 Rg2 nL2

where Rg is the average square radius of gyration experimentally accessible. This relation is considered extremely important in the sense that any conformational perturbation is amplified by a factor n in the final value of C∞ (or of ), and therefore represents a highly demanding test for the appropriateness of the conformational calculations and at the same time a discriminating factor for conformation-dependent solution properties (e.g., viscosity). An alternative measure of chain extension is the persistent length, Pn, which behaves in a similar asymptotic dependence on n as observed for Cn, in Figure 12.2.11. It is meant as the capacity of the chain to preserve the direction of the first residue (vector) of the chain. Therefore, as the directional persistence dissipates with Figure 12.2.11. Characteristic ratio, correlation function increasing chain length and the direction of and persistent length as a function of the degree of poly- the terminal residue vector Ln loses correlamerization, n, for [α-D-(1-3)-glc]n (pseudonigeran) and tion with that of the initial vector L1, Pn ap(b) the [β-D-(1-3)-glc]n (curdlan) calculated on the in proaches an asymptotic limit for vacuo energy maps. sufficiently long chains. Both the Cn and Pn functions in Figure 12.2.11 reveal the [α-D-(1-3)-glc]n chain to be much more extended than that of [β-D-(1-3)-glc]n. The directional correlations are also well characterized by the correlation function Fn of equations [12.2.6], which measures the average projection of a unit vector aligned with each virtual bond of the chain onto the unit vector relative to the first residue. The strongly oscillating character of Fn for [β-D-(1-3)-glc]n, which is also observed at low n in both the Cn and Pn functions, reflects the pseudo-helical persistence of the backbone trajectory which becomes uncorrelated (i.e., Fn declines to zero) as the molecular weight increases. It is noteworthy that the oscillations in Fn retain approximately the pseudo-helical periodicity present in the crystalline forms of that polysaccharide (Figure 12.2.2). The monotonic decline of Fn (a) for [α-D-(1-3)-glc]n, on the other hand, shows that this polymer possess a stronger directional correlation due to the different glycosidic linkage which has a dramatic effect on the character of the chain trajectory. The chain properties illustrated above are often described with terms such as “stiffness” and “flexibility”. It is important to clarify that stiffness and structural rigidity may not necessarily be alternative to flexibility and structural disorder, since two different concepts enter into the above definitions: one is concerned with the number of different accessible conformations, the other with the (average) direction of the sequential bonds, i.e., with the chain topology.


725

Figure 12.2.12. Characteristic ratio, Cn, as a function of degree of polymerization, n, for amylose (a) and cellulose (b) in vacuo, water and DMSO.

By calculating of the conformational energy surface of a dimer as a starting point for the prediction of mean chain properties, the effect of the solvation on the chain conformation is taken into account by evaluation of the perturbations of such surface due to the presence of the solvent. The characteristic ratio (eq. [12.2.6]) has been computed for (a) [β-D-(1-4)-glc]n (cellulose) and (b) [α-D-(1-4)-glc]n (amylose) in different solvents, namely, water and DMSO.53 The results are reported in Figure 12.2.12 where, for comparison, the in vacuo Cn for both polymers is shown. Figure 12.2.12a shows that the chain extension of cellulose decreases significantly in DMSO and much more so in water with a reduction of about 50% when compared to that in vacuo data (C∞=96). The solvation contribution of water seems to favor those conformations, which have smaller cavity volumes and thus force the chain to minor extensions. Tanner and Berry66 reported, for cellulose derivatives in solution, a value for the limiting characteristic ratio of between 30 and 60, which is in good agreement with the theoretical data in Figure 12.2.12a. Figure 12.2.12b shows the characteristic ratio of amylose chains computed in the same solvents as in Figure 12.2.12a. The amylose chain extension is considerably lower than that of cellulose and, as in the case of [β-D-(1-3)-glc]n shown above, presents at low n a remark-

726


able pseudo-helical pattern. With respect to cellulose, the amylose chain shows a more coiled and apparently disordered backbone topology, but from a statistical point of view possesses a lower configurational entropy,67 i.e., a more limited set of allowed conformational states. The C∞ values are higher in water (5.6) than in DMSO (4.9) and both are higher than for the unperturbed chain which is in good agreement with the earlier work of Jordan and Brant67 which observed a decrease of about 20% in chain dimensions of amylose DMSO/water mixture with respect to the water alone. More recently, Nakanishi and co-workers68 have demonstrated from light scattering, sedimentation equilibrium and viscosity measurements on narrow distribution samples, that the amylose chain conformation in DMSO is a random coil, which results expanded (C∞ = 5) by excluded-volume effect at high molecular weight. Norisuye69 elaborated a set of published viscosity data reporting a C∞ = 4.2-4.5 for unperturbed amylose and C∞ = 5.3 for aqueous KCl solutions, while Ring and co-workers70 estimated a C∞ ≅ 4.5 for amylose/water solutions by means of the Orofino-Flory theory of the second virial coefficient. Order-disorder conformational transitions very often occur on changing physical and/or chemical conditions of polysaccharide solutions. DMSO, for example, is the solvent, which is commonly used as co-solvent for stabilizing or destabilizing ordered solution conformations. Schizophyllan, a triple helical polysaccharide with a [β-D-(1-3)-glc]n backbone exhibits a highly cooperative order-disorder transition in aqueous solution.71 When small quantities of DMSO are added to aqueous solutions the ordered state is remarkably stabilized, as has been observed in the heat capacity curves by means of the DSC technique.71 Several efforts have been made with MD simulations in order to explicitly take into account the solvent molecule effect on the saccharide conformation, although only oligomeric segments have been considered, given the complexity in terms of computational time required for such a multi-atoms system. One interesting example is that of Brady and co-workers56 on the stability and the behavior of double-helix carrageenan oligomer in aqueous solution compared with the results of the in vacuo calculations. They observed a higher relative stability of the double helix in vacuo, a fact, which is consistent with experimental results under anhydrous conditions, as in the fiber diffraction studies. However, in aqueous solution, the interchain hydrogen bonds that stabilize the double-helix structure appear much less stable, as the glycosidic hydroxyl groups make more favorable interactions with water molecules. They concluded that in the solvation step the double-helix would seem to be unstable and an unwinding process is theoretically predicted, at least for the oligomers. 12.2.6 SOLVENT EFFECT ON CHARGED POLYSACCHARIDES AND THE POLYELECTROLYTE MODEL 12.2.6.1 Experimental behavior of polysaccharides polyelectrolytes Based on the experimental evidence of polyelectrolyte solutions, whenever the degree of polymerization is sufficiently high, all ionic macromolecules are characterized by a peculiar behavior, which sets them apart from all other ionic low molecular weight molecules as well as from non-ionic macromolecules. A general consequence of the presence of charged groups in a chain is a favorable contribution to the solubility of polymer in water. A strongly attractive potential is generated between the charge density on the polymer and the opposite charges in solution. For example, the value of the activity coefficient of the counterions is strongly reduced with respect to that of the same ions in the presence of the univalent opposite charged species. If the charge density of the polyelectrolyte is sufficiently high, such a


727

phenomenon is justified through a ‘condensation process’ of counterions and it has also been interpreted theoretically. On the polymer side, among the dramatic changes that the presence of charged groups imparts to solution properties, there are the enhanced chain dimensions, the increased hydrodynamic volume (i.e., viscosity), and, in general, a strong influence on all conformational properties. Subject to the constraints imposed by the chemical structure of the chain, the distribution of charged groups and their degree of ionization contribute to determining the equilibrium chain conformation; both the Coulombic interaction among the charged groups and the distribution and concentration of the screening counterions are important. Most of the physico-chemical properties of the system result from a non-linear combination of these parameters. However, one has not to forget that the variability of conformation alters the distances between charged groups on the polymeric chain and that the equilibrium is statistically defined by the Gibbs energy minimum of the system. As an important consequence of this energy balance, changes in temperature, ionic strength, pH, etc., can provoke changes in polyelectrolyte conformation, often cooperatively in the case of biopolymers, between states with different values of the charge density. These states may be characterized by different structural orders (e.g. helix →extended chain transition), by different degrees of flexibility of the chain (globular coil → expanded chain) or by different extent of aggregation (monomeric → dimeric or multimeric chains). Theoretical calculations based on molecular grounds are still extremely complicated and incomplete72 and other routes must be more empirically used in order to interpret the experimental data and to understand the correlation between conformational properties and structure. The central problem is to quantify the interactions among charges on the polymer and among these same charges and their respective counterions. As far as it concerns the short-range interactions, the introduction of charged groups modifies the equilibrium geometry of the monomeric units and the contribution of the electrostatic nature on the nearest-neighbor conformational energy. These conclusions also derive from the already demonstrated effect of the solvent interactions on the unperturbed dimensions of amylose and cellulose,53 and from the evidence of the perturbation on the conformational energy surface of several charged saccharidic units.73 There at least two approaches that may be relevant for this review; one is that described by Haug and Smidsrød74 for the rationalization of the dimensional properties of polyelectrolytes as a function of salt concentration, the other is the formulation of a statistical thermodynamic theory for the “physical” framing of the ion-polyelectrolyte interactions. Both these theoretical formulations deal with the conformation of the polymer and predict that the conformational features must be function of ionic strength (see for example refs. 75 and 76). 12.2.6.2 The Haug and Smidsrød parameter: description of the salt effect on the chain dimension A peculiarity of the correlation between the viscometric parameters and the dimensions of the macromolecular chain has long been recognized and theoretical approaches have been developed for several chain models.77,78 The behavior of polyelectrolytes adds some complications especially in the low ionic strength regime. It has however been understood that the intrinsic viscosity, [η], of a polyion (i.e., its hydrodynamic volume) decreases with increasing ionic strength, I, as a consequence of the screening of the fixed charges on the

728


Figure 12.2.13. Dependence of the intrinsic viscosity [η] of hyaluronic acid (ξ=0.72) and its benzyl derivatives with decreasing linear charge density, ξ, on the inverse square root of ionic strength, I, at 25°C.

polyion. At infinite ionic strength, the chain dimensions may eventually correspond to the completely uncharged macromolecule. This is a sort of “ideal state” of the polyelectrolyte; “ideal” with respect to the long-range electrostatic repulsive interactions only, without relation to the Θ-conditions. In the absence of a cooperative conformational transition, for many polysaccharide polyelectrolytes a linear dependence of [η] upon I1/2 is observed, with the slope diminishing with the charge density associated to the polysaccharide chain (Figure 12.2.13). A theory has been presented for an estimation of the relative stiffness of the molecular chains by Smidsrød and Haug,74 which is based on the Fixman’s theory and Mark-Houwink equation. The chain stiffness parameter is estimated from the normalized slope B of [η] vs. the inverse square root of the ionic strength: ∂[η] ∂I −1/ 2

= slope = B([η]0.1 )

γ

[12.2.7]

where: γ

has a value between 1.2 and 1.4

The dependence of viscosity on the ionic strength, as given in equation above, has been increasingly popular in the field of polysaccharides with the purpose of comparing the chain stiffness of different macromolecules. The derivation is based on the Fixman theory, which defines the dependence of [η] on the molecular weight through an expansion coefficient which effectively takes into account the electrostatic interactions in the Debye-Hückel approximation. The semi-empirical treatment of the hydrodynamic properties of statistical polyelectrolytes (at sufficiently high values of the ionic strength) is built upon a straightforward extension of the theory of intermolecular interactions for uncharged polymers, for which a linear relation can be written between the expansion coefficient, α 3η , and the square root of the molecular weight, M. It should also be mentioned78 that the various theoretical treatments of the salt dependence of the excluded volume and of the expansion coefficients


729

led to a linearity of α 3η on (M/Cs)1/2 only over a limited range of salt concentration. On varying the salt concentration, one effectively deals with a set of binary solvents with a variety of interaction parameters. The approach proposed by Smidsrød overcomes the indetermination of some parameters by using the slope of [η] as a function of the inverse square-root of the ionic strength of the medium. The ultimate relationship is obtained between the constant B and the effective bond length b θ (B = const·b θ -2), which, in the absence of any reasonable knowledge of the constant, can only be used in an empirical way. There is a compelling although intuitive limit to the use of the Smidsrød-Haug approach for those macroions that do not counterbalance the effective electrostatic field exerted by the ionic strength through the conformational elasticity. Among these polymers, those characterized by a low value of fixed charges, of flexibility, and/or of molecular weight fall behind the limits of the correct applicability of the Smidsrød-Haug approach, which should maintain its validity only for gaussian chains with prominent electrostatic interactions. Most likely, the Smidsrød-Haug parameter has been abused in the field of polysaccharides without the authors’ intention, in the sense that the original treatment was aware about the intrinsic limitations of the approach,74 while the extensions have thereafter been considered as permitted. We wish to point out that the above comment does not imply the failure of the linearity of [η] with I-1/2, but only a meaningless result for the values of b θ obtained for low-charged polysaccharide polyelectrolytes. Reference can be made to two series of polysaccharides (chitosans and hyaluronans), which have been investigated in some detail for the specific application of the Smidsrød-Haug approach.79,80 12.2.6.3 The statistical thermodynamic counterion-condensation theory of Manning Linear polyelectrolytes bear a charge distribution along the chain, properly neutralized by small ions of opposite sign. In the absence of added salt, it is reasonably assumed that the charge density (when sufficiently high) will increase the local stiffness of the chain because of the electrostatic repulsion. For this reason the linear polyelectrolyte is often regarded as a charged rod. Osawa has introduced the concept of a critical charge density on the polymer and described the ion- pairing of polyion and counterions as condensation. The model described here has been extensively used in previous papers of the authors81 and described in the original papers by Manning.82 Let here simply give a few comments on the physical basis of this model relevant to the present case. The counterion condensation (CC) theory, largely developed by Manning, gives analytical solutions to evaluate the electrostatic potential around a linear polyelectrolyte, provided its conformation is regular and fixed. The rigidity of the polymer seems therefore to be both a prerequisite and a result of the molecular polyelectrolyte theory, and it has been thought to be not too far from reality in many cases.82 The application of the above theory to experimental results has been carried out with the assumption that the charge distribution is structurally defined by the monomer repeat as derived from the solid state fiber diffractograms, although sometimes the fully stretched chain conformation has been taken (Figure 12.2.14). The original work of Manning’s counterion condensation theory has provided an elegant tool for describing several properties of polyelectrolytes in terms of the structural parameter, ξ, which is unequivocally defined as e2/εkTb, where e is the value of the elementary charge, ε is the dielectric constant of the medium, k is the Boltzmann constant, T is the Kelvin temperature, and b is the distance between the projections of the fixed charges of the polyelectrolyte on its contour axis. In the case of monovalent ions, for all

730


Figure 12.2.14. The rigid chain (a), the corresponding polyelectrolyte model (b) and the realistic flexible model of a polysaccharidic chain (c).

structural values of ξ> 1, a defined amount of counterions will “condense” from the solution into the domain of the polymer chain so as to reduce the “effective” value of ξ to unity. For water, ξ = 0.714/b (with b expressed in nm), and is practically independent of T, being the electrostatic-excess Gibbs free energy of the solution given by: G el = −ξln[1 − exp(−Kb )] where K is the Debye-Hückel screening parameter. Application of this theory has been made to many experimental cases, and in particular an extensive correlation has been made between the theoretical predictions and the thermodynamic data on the processes of protonation, dilution and mixing with ions.76 Besides these nice applications of the theory to problems with a strong “academic” character, there is another very striking example of prediction of solvent-induced conformational changes for the effects of salts on the conformational stability of ordered polyelectrolytes. In fact, in addition to the condensation phenomena predicted by the polyelectrolitic theory, other physical responses may also occur (also simultaneously), which may mask this central statement of the Manning theory “that the onset of the critical value ξ (for univalent ions ξ > 1) constitutes a thermodynamic instability which must be compensated by counterion condensation”. In fact, chain extension and/or disaggregation of aggregated chains may occur or change upon the variation of charge density, and the energetic instability effectively becomes a function of the thermodynamic state of the polyelectrolytic chain. The range of theoretical and experimental approaches has been, in particular, addressed to the problem of conformational transitions between two different states, provided they have different charge densities. For thermally induced, conformational transitions be-


731

tween states i and f of a polyelectrolyte, characterized by a set of ξi and ξf (i.e., bi and bf) values, polyelectrolyte theory predicts a simple relationship82 between the values of the melting temperatures (TM, the temperature of transition midpoint) and the logarithm of the ionic strength, I:

( ) = − 9.575F (ξ)

d TM−1

d (log I)

∆M H

where ∆ M H is the value of the enthalpy of transition (in J per mole of charged groups) determined calorimetrically. This linearity implies, indeed, that the enthalpy change is essentially due to non-ionic contributions and largely independent of I. The function F(ξ) depends on the charge density of both the final state (subscript f) and the initial state (subscript i), within the common condition that ξ f < ξ i, that is the final state is characterized by a smaller value of the charge density. The value of F(ξ) is given in the literature. This relation has been successfully applied first to the transition processes of DNA,83 polynucleotides,84 but also to many ionic polysaccharides (carrageenans,85 xanthan,86 succinoglycan,87) of great industrial interest. Accurate determination of the TM values of the polysaccharide as a function of the ionic strength is necessary. 12.2.6.4 Conformational calculations of charged polysaccharides The major problem for conformational calculations of ionic polysaccharides arises from the correct evaluation of the electrostatic potential energy due to the charged groups along the chain and to the all other ions in solutions. The interaction between the polyion charges and the counterions is formally non-conformational but it largely affects the distribution of the conformational states. Ionic polymers are often simplistically treated either in the approximation of full screening of the charged groups or in the approximation of rigid conformational states (regular rod-like polyelectrolyte models). A combination of the molecular polyelectrolyte theory82,83 with the methods of statistical mechanics can be used at least for the description of the chain expansion due to charges along the polysaccharide chain. The physical process of the proton dissociation of a (weak) polyacid is a good way to assess the conformational role of the polyelectrolytic interactions, since it is possible of tuning polyelectrolyte charge density on an otherwise constant chemical structure. An amylose chain, selectively oxidized on carbon 6 to produce a carboxylic (uronic) group, has proved to be a good example to test theoretical results.81 If the real semi-flexible chain of infinite length is replaced by a sequence of segments, the average end-to-end distance of each segment defines the average distance between charges: b =

r N

[12.2.8]

where N is the number of charges in the segment. The distance between charges fluctuates within the limits of the conformational flexibility of the chain, as calculated by the proper non-bonding inter-residue interactions. The probability function W'(r) of the end-to-end displacement r of a charged segment can be obtained by multiplying its a priori (non-ionic) probability W(r) with the Boltzmann

732


term involving the excess electrostatic free energy (Figure 12.2.15). The probability theory guarantees both that the components (repeating units) of the segment vectors be distributed in a Gaussian way along the chain segment, and that high molecular weight polymers be composed by a statistical sequence of those segments. Consequence of the above approximation is that the distance r between any two points of the chain (separated by a sufficiently large number of residues, n) does not depend on the specific sequence and values of conformational angles and energies, but only upon the average potential summed over the number of residues n. The calculation of the averaged (electrostatic) functions is reached in two steps. At the first, the proper flexibility of the Figure 12.2.15. Dependence of the probability distribuis evaluated either from tion function of a model semi-flexible chain, (a) un- polymer charged and charged (a'), on the end-to-end distance and conformational calculation or from suitable the respective total free energies (c and c'). The electro- models, then the mean value of each propstatic contribution (b) is also reported. erty is calculated through the averaging procedure described below. The computational procedure is the following: • the conformational energy surface of the uncharged polymer is evaluated by the standard methods the conformational analysis;65 • the end-to-end distribution distance Wn(r) for the (uncharged) polymer segments is determined by numerical Monte Carlo methods;64 • the dependence of the total (conformational) energy G(r) upon chain extension r is therefore estimated from the distribution of segment lengths; a Boltzmannian distribution is assumed. In most cases the distribution function is Gaussian (or approximately so) and the corresponding free energy function can be approximated by a simple parabolic equation (Figure 12.2.15). In this case, we assume a Hookean energy (which is correct at least for the region around the maximum of the distribution curve), so we have:  G  0 W (r ) = A exp  − , G(r ) = k r − r  RT 

(

)

[12.2.9]

where: r0 k

average segment length a constant which determines the flexibility of the chain

The ionic energy, that results from the process of charging the polymer groups, changes the probability of the end-to-end distance for the i-th segment, W'(r), to the probability of the average inter-charge separation distance , W(b), following the definition of equation [12.2.8] and [12.2.9].


733

The conformational (non-ionic) free energy, obtained from the radial distribution function for non-ionic chains by Monte Carlo calculations, was used in conjunction with the electrostatic free energy to calculate the actual distribution function of the charged chain segments. The resulting expansion justifies almost quantitatively in many cases the experimental thermodynamic properties (such as pKa, Hdil, etc.) and the dimensional properties (viscosity) of the ionic polysaccharides to which the approach has been applied. 12.2.7 CONCLUSIONS Only some aspects of the solvent perturbation on the conformational properties of carbohydrate polymers have been covered in this chapter. One of the major concerns has been to develop a description of these “solvent effects” starting with the complex conformational equilibria of simple sugars. In fact, only recently it has been fully appreciated the quantitative relationship between conformational population and physical properties, e.g. optical rotation. The chapter, however, does not give extensive references to the experimental determination of the polysaccharide shape and size in different solvents, but rather it attempts to focus on the molecular reasons of these perturbations. A digression is also made to include the electrostatic charges in polyelectrolytic polysaccharides, because of their diffusion and use and because of interesting variations occurring in these systems. Thus, provided that all the interactions are taken into account, the calculation of the energetic state of each conformation provides the quantitative definition of the chain dimensions. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

J.R. Brisson and J.P. Carver, Biochemistry, 22, 3671 (1983). R.C. Hughes and N. Sharon, Nature, 274, 637 (1978). D.A. Brant, Q. Rev. Biophys., 9, 527 (1976). B.A. Burton and D.A. Brant, Biopolymers, 22, 1769 (1983). G.S. Buliga and D.A. Brant, Int. J. Biol. Macromol., 9, 71 (1987). V. S. R. Rao, P. K. Qasba, P. V. Balaji and R. Chandrasekaran, Conformation of Carbohydrates, Harwood Academic Publ., Amsterdam, 1998, and references therein. R. H. Marchessault and Y. Deslandes, Carbohydr. Polymers, 1, 31 (1981). D. A. Brant, Carbohydr. Polymers, 2, 232 (1982). P.R. Straub and D.A. Brant, Biopolymers, 19, 639 (1980). A. Cesàro in Thermodynamic Data for Biochemistry and Biotechnology, H.J. Hinz (Ed.), Springer-Verlag, Berlin, 1986, pp. 177-207. Q. Liu and J.W. Brady, J. Phys.Chem. B, 101, 1317 (1997). J.W. Brady, Curr. Opin. Struct.Biol., 1, 711 (1991). Q. Liu and J.W. Brady, J. Am. Chem. Soc., 118, 12276 (1996). I. Tvaroška, Biopolymers, 21, 188 (1982). I. Tvaroška, Curr. Opin. Struct.Biol., 2, 661 (1991). K. Mazeau and I. Tvaroška, Carbohydr. Res., 225, 27 (1992). Perico, A., Mormino, M., Urbani, R., Cesàro, A., Tylianakis, E., Dais, P. and Brant, D. A., Phys. Chem. B, 103, 8162-8171(1999). K.D. Goebel, C.E. Harvie and D.A. Brant, Appl. Polym. Symp., 28, 671 (1976). M. Ragazzi, D.R. Ferro, B. Perly, G. Torri, B. Casu, P. Sinay, M. Petitou and J. Choay, Carbohydr. Res., 165, C1 (1987). P.E. Marszalek, A.F. Oberhauser, Y.-P. Pang and J.M. Fernandez, Nature, 396, 661 (1998). S.J. Angyal, Aust. J.Chem., 21, 2737 (1968). S.J. Angyal, Advan. Carbohyd. Chem. Biochem., 49, 35 (1991). D.A. Brant and M.D. Christ in Computer Modeling of Carbohydrate Molecules, A.D. French and J.W. Brady, Eds., ACS Symposium Series 430, ACS, Washington, DC, 1990, pp. 42-68. R. Harris, T.J. Rutherford, M.J. Milton and S.W. Homans, J. Biomol. NMR, 9, 47 (1997). M. Kadkhodaei and D.A. Brant, Macromolecules, 31, 1581 (1991).

734


I. Tvaroška and J. Gajdos, Carbohydr. Res., 271, 151 (1995). F.R. Taravel , K. Mazeau and I. Tvaroška, Biol. Res., 28, 723 (1995). J.L. Asensio and J. Jimenez-Barbero, Biopolymers, 35, 55 (1995). P. Dais, Advan. Carbohyd. Chem. Biochem., 51, 63 (1995). F. Cavalieri, E. Chiessi, M. Paci, G. Paradossi, A. Flaibani and A. Cesàro, Macromolecules, (submitted). I. Tvaroška and F.R. Taravel, J. Biomol. NMR, 2, 421 (1992). R.E. Wasylishen and T. Shaefer, Can. J. Chem., 51, 961 (1973). A.A. van Beuzekom, F.A.A.M. de Leeuw and C. Altona, Magn. Reson. Chem., 28, 888 (1990). I. Tvaroška, Carbohydr. Res., 206, 55 (1990). I. Tvaroška and F.R. Taravel, Carbohydr. Res., 221, 83 (1991). J.P. Carver, D. Mandel, S.W. Michnick, A. Imberty, J.W. Brady in Computer Modeling of Carbohydrate Molecules, A.D. French and J.W. Brady, Eds., ACS Symposium Series 430, ACS, Washington, DC, 1990, pp. 267-280. 37 D. Rees and D. Thom, J. Chem. Soc. Perkin Trans. II, 191 (1977). 38 E.S. Stevens, Carbohydr. Res., 244, 191 (1993). 39 D.H. Whiffen, Chem. Ind., 964 (1956). 40 J.H. Brewster, J. Am. Chem. Soc., 81, 5483 (1959). 41 D. A. Rees, J. Chem. Soc. B, 877 (1970). 42 E.S. Stevens and B.K. Sathyanarayana, J.Am.Chem.Soc., 111, 4149 (1989). 43 C.A. Duda and E.S. Stevens, Carbohydr. Res., 206, 347 (1990). 44 R. Urbani, A. Di Blas and A. Cesàro, Int. J. Biol. Macromol., 15, 24 (1993). 45. I. Tvaroška, S. Perez, O. Noble and F. Taravel, Biopolymers, 26, 1499 (1987). 46 C. Gouvion, K. Mazeau, A. Heyraud, F. Taravel, Carbohydr. Res., 261, 261 (1994). 47 R.A. Pierotti, Chem. Rev., 76, 717 (1976). 48 R.J. Abraham and E. Bretschneider in Internal Rotation in Molecules, W.J. Orville-Thomas, Ed., Academic Press, London, 1974, ch. 13. 49 D.L. Beveridge, M.M. Kelly and R.J. Radna, J. Am. Chem. Soc., 96, 3769 (1974). 50 M. Irisa, K. Nagayama and F. Hirata, Chem. Phys. Letters, 207, 430 (1993). 51 R.R. Birge, M.J. Sullivan and B.J. Kohler, J. Am. Chem.Soc., 98, 358 (1976). 52 I. Tvaroška and T. Kozar, J. Am. Chem. Soc., 102, 6929 (1980). 53 R. Urbani and A. Cesàro, Polymers, 32, 3013 (1991). 54 J.W. Brady, J. Am. Chem. Soc., 108, 8153 (1986). 55 C.B. Post, B.R. Brooks, M. Karplus, C.M. Dobson, P.J. Artymiuk, J.C. Cheetham and D.C. Phillips, J. Mol. Biol., 190, 455 (1986). 56 K. Ueda, A. Imamura and J.W. Brady, J. Phys. Chem., 102, 2749 (1998). 57 R. Grigera, Advan. Comp. Biol., 1, 203 (1994). 58 S.N. Ha, A. Giammona, M. Field and J.W. Brady, Carbohydr. Res., 180, 207 (1988). 59 C.L. Brooks, M. Karplus, B.M. Pettitt, Proteins: A Theoretical Perspective of Dynamics, Structure and Thermodynamics, Wiley Interscience, New York, 1988, vol. LXXI. 60 R. Schmidt, B. Teo, M. Karplus and J.W. Brady, J. Am. Chem. Soc., 118, 541 (1996). 61 Q. Liu, R. Schmidt, B. Teo, P.A. Karplus and J.W. Brady, J. Am. Chem. Soc., 119, 7851 (1997). 62 V.H. Tran and J.W. Brady, Biopolymers, 29, 977 (1990). 63 M. Karplus and J.N. Kushick, Macromolecules, 31, 1581 (1981). 64 R.C. Jordan, D.A. Brant and A. Cesàro, Biopolymers, 17, 2617 (1978). 65 P.J. Flory, Statistical Mechanics of Chain Molecules, Wiley-Interscience, New York, 1969. 66 D.W. Tanner and G.C. Berry, J. Polym. Sci., Polym. Phys. Edn., 12, 441 (1974). 67 R.C. Jordan and D.A. Brant, Macromolecules, 13, 491 (1980). 68 Y. Nakanishi, T. Norisuye, A. Teramoto and S. Kitamura, Macromolecules, 26, 4220 (1993). 69 T. Norisuye, Food Hydrocoll., 10, 109 (1996). 70 S.G. Ring, K.J. Anson and V.J. Morris, Macromolecules, 18, 182 (1985). 71 T. Hirao, T. Sato, A. Teramoto, T. Matsuo and H. Suga, Biopolymers, 29, 1867 (1990). 72 C. Sagui and T.A. Darden, Ann. Rev. Biophys. Biomol. Struct., 28, 155 (1999). 73 J.R. Ruggiero, R. Urbani and A. Cesàro, Int. J. Biol. Macromol., 17, 205 (1995). 74 O. Smidsrød and A. Haug, Biopolymers, 10, 1213 (1971). 75 G.S. Manning and S. Paoletti, in Industrial Polysaccharides, V. Crescenzi, I.C.M. Dea and S.S. Stivala, Eds., Gordon & Breach, N Y, 1987, pp. 305-324. 76 S. Paoletti, A. Cesàro, F. Delben, V. Crescenzi and R. Rizzo, in Microdomains in Polymer Solutions, P. Dubin Ed., Plenum Press, New York, (1985) pp 159-189. 26 27 28 29 30 31 32 33 34 35 36


77 78 79 80 81 82 83 84 85 86 87

735

H. Morawetz, Macromolecules in Solution, Interscience, New York, 1975, Ch. 7. M. Bohdanecký and J. Kovár, Viscosity of Polymer Solutions, Elsevier, Amsterdam, 1982, p. 108. M.W. Anthonsen, K.M. Vårum and O. Smidsrød, Carbohydr. Polymers, 22, 193 (1993). R. Geciova, A. Flaibani, F. Delben, G. Liut, R. Urbani and A. Cesàro, Macromol. Chem. Phys., 196, 2891 (1995). A. Cesàro, S. Paoletti, R. Urbani and J.C. Benegas, Int. J. Biol. Macromol., 11, 66 (1989). G.S. Manning, Acc. Chem. Res., 12, 443 (1979). G.S. Manning, Quart. Rev. Biophys., 11, 179 (1978). M.T. Record, C.F. Anderson and T.M. Lohman, Quart. Rev. Biophys., 11, 103 (1978). S. Paoletti, F. Delben, A. Cesàro and H. Grasdalen, Macromolecules, 18, 1834 (1985). S. Paoletti, A. Cesàro and F. Delben, Carbohydr. Res., 123, 173 (1983). T.V. Burova, I.A. Golubeva, N.V. Grinberg, A.Ya. Mashkevich, V.Ya. Grinberg, A.I.Usov, L. Navarini and A. Cesàro, Biopolymers, 39, 517 (1996).

13

Effect of Solvent on Chemical Reactions and Reactivity 13.1 SOLVENT EFFECTS ON CHEMICAL REACTIVITY Roland Schmid Technical University of Vienna Institute of Inorganic Chemistry, Vienna, Austria

13.1.1 INTRODUCTION About a century ago, it was discovered that the solvent can dramatically change the rate of chemical reactions.1 Since then, the generality and importance of solvent effects on chemical reactivity (rate constants or equilibrium constants) has been widely acknowledged. It can be said without much exaggeration that studying solvent effects is one of the most central topics of chemistry and remains ever-increasingly active. In the course of development, there are few topics in chemistry in which so many controversies and changes in interpretation have arisen as in the issue of characterizing solute-solvent interactions. In a historical context, two basic approaches to treating solvent effects may be distinguished: a phenomenological approach and a physical approach. The former may be subdivided further into the dielectric approach and the chemical approach. • Phenomenological approach Dielectric Chemical • Physical approach That what follows is not intended just to give an overview of existing ideas, but instead to filter seminal conceptions and to take up more fundamental ideas. It should be mentioned that solvent relaxation phenomena, i.e., dynamic solvent effects, are omitted. 13.1.2 THE DIELECTRIC APPROACH It has soon been found that solvent effects are particularly large for reactions in which charge is either developed or localized or vice versa, that is, disappearance of charge or spreading out of charge. In the framework of electrostatic considerations, which have been around since Berzelius, these observations led to the concept of solvation. Weak electrostatic interactions simply created a loose solvation shell around a solute molecule. It was in this climate of opinion that Hughes and Ingold2 presented the first satisfactory qualitative account of solvent effects on reactivity by the concept of activated complex solvation.

738

Roland Schmid

The first solvent property applied to correlate reactivity data was the static dielectric constant ε (also termed εs) in the form of dielectric functions as suggested from elementary electrostatic theories as those by Born (1/ε), Kirkwood (ε-1)/(2ε+1), ClausiusMosotti (ε-1)/(ε+2), and (ε-1)/(ε+1). A successful correlation is shown in Figure 13.1.1 for the rate of the SN2 reaction of p-nitrofluorobenzene with piperidine.3 The classical dielecFigure 13.1.1. Relationship between second-order rate constants at tric functions predict that reactiv50°C of the reaction of p-nitrofluorobenzene and piperidine and the ity changes level out for dielectric solvent dielectric properties [from ref. 21]. constants say above 30. For instance, the Kirkwood function has an upper limiting value of 0.5, with the value of 0.47 reached at ε = 25. The insert in Figure 13.1.1 illustrates this point. Therefore, since it has no limiting value, the log ε function may be preferred. A theoretical justification can be given in the framework of the dielectric saturation model of Block and Walker.4 Picturing the solvent as a homogeneous dielectric continuum means in essence that the solvent molecules have zero size and that the molecules cannot move. The most adequate physical realization would be a lattice of permanent point dipoles that can rotate but cannot translate. 13.1.3 THE CHEMICAL APPROACH Because of the often-observed inadequacies of the dielectric approach, that is, using the dielectric constant to order reactivity changes, the problem of correlating solvent effects was next tackled by the use of empirical solvent parameters measuring some solvent-sensitive physical property of a solute chosen as the model compound. Of these, spectral properties such as solvatochromic and NMR shifts have made a spectacular contribution. Other important scales are based on enthalpy data, with the best-known example being the donor number (DN) measuring solvent’s Lewis basicity. In the intervening years there is a proliferation of solvent scales that is really alarming. It was the merit particularly of Gutmann and his group to disentangle the great body of empirical parameters on the basis of the famous donor-acceptor concept or the coordination-chemical approach.5 This concept has its roots in the ideas of Lewis going back to 1923, with the terms donor and acceptor introduced by Sidgwick.6 In this framework, the two outstanding properties of a solvent are its donor (nucleophilic, basic, cation-solvating) and acceptor (electrophilic, acidic, anion-solvating) abilities, and solute-solvent interactions are considered as acid-base reactions in the Lewis’ sense. Actually, many empirical parameters can be lumped into two broad classes, as judged from the rough interrelationships found between various scales.7 The one class is more concerned with cation (or positive dipole’s end) solvation, with the most popular solvent basic-

13.1 Solvent effects on chemical reactivity

739

ity scales being the Gutmann DN, the Kamlet and Taft β, and the Koppel and Palm B. The other class is said to reflect anion (or negative dipole’s end) solvation. This latter class includes the famous scales π*, α, ET(30), Z, and last but not least, the acceptor number AN. Summed up: Cation (or positive dipole’s end) solvation DN • Gutmann • Kamlet and Taft β B (B*) • Koppel and Palm Anion (and negative dipole’s end) solvation Gutmann AN Dimroth and Reichardt ET(30) Kosower Z Kamlet and Taft α, π* These two sets of scales agree in their general trend, but are often at variance when values for any two particular solvents are taken. Some intercorrelations have been presented by Taft et al., e.g., the parameters ET, AN and Z can be written as linear functions of both α and π*.8 Originally, the values of ET and π* were conceived as microscopic polarity scales reflecting the “local” polarity of the solvent in the neighborhood of solutes (“effective” dielectric constant in contrast to the macroscopic one). In the framework of the donor-acceptor concept, however, they obtained an alternative meaning, based on the interrelationships found between various scales. Along these lines, the common solvents may be separated into six classes as follows. 1 nonpolar aliphatic solvents 2 protics or protogenetic solvents (at least one hydrogen atom is bonded to oxygen) 3 aromatic solvents 4 (poly)halogenated solvents 5 (perhaps) amines 6 select (or “normal” according to Abraham) solvents defined as non-protonic, non-chlorinated, aliphatic solvents with a single dominant bond dipole. A case study is the plot of AN versus ET shown in Figure 13.1.2. While there is a quite good correspondence for the select solvents (and likely for the nonpolar aliphatic solvents), the other classes are considerably off-line.9 This behavior may be interpreted in terms of the operation of different solvation mechanisms such as electronic polarizability, dipole density, and/or hydrogen-bonding (HB) ability. For instance, the Figure 13.1.2. Relationship between the ET(30) values and the ac- main physical difference between ceptor number [from ref. 21]. Triangles: protic solvents, squares: ar- π* and ET(30), in the absence of

• • • •

omatic and chlorinated solvents.

740

Roland Schmid

HB interactions, is claimed to lie in different responses to solvent polarizability effects. Likewise, in the relationship between the π* scale and the reaction field functions of the refractive index (whose square is called the optical dielectric constant e∞) and the dielectric constant, the aromatic and the halogenated solvents were found to constitute special cases.10 This feature is also reflected by the polarizability correction term in eq. [13.1.2] below. For the select solvents, the various “polarity” scales are more or less equivalent. A recent account of the various scales has been given by Marcus,11 and in particular of π* by Laurence et al.,12 and of ET by Reichardt.13 However, solvation is not the only mode of action taken by the solvent on chemical reactivity. Since chemical reactions typically are accompanied by changes in volume, even reactions with no alteration of charge distribution are sensitive to the solvent. The solvent dependence of a reaction where both reactants and products are neutral species (“neutral” pathway) is often treated in terms of either of two solvent properties. The one is the cohesive energy density ε c or cohesive pressure measuring the total molecular cohesion per unit volume, ε c = (∆Hv − RT ) / V

[13.1.1]

where: ∆Hv V

molar enthalpy of vaporization molar liquid volume

The square root of ε c is termed the Hildebrand solubility parameter δH, which is the solvent property that measures the work necessary to separate the solvent molecules (disrupt and reorganize solvent/solvent interactions) to create a suitably sized cavity for the solute. The other quantity in use is the internal pressure Pi which is a measure of the change in internal energy U of the solvent during a small isothermal expansion, Pi = (∂U/∂V)T. Interesting, and long-known, is the fact that for the highly dipolar and particular for the protic solvents, values of ε c are far in excess of Pi.14 This is interpreted to mean that a small expansion does not disrupt all of the intermolecular interactions associated with the liquid state. It has been suggested that Pi does not detect hydrogen bonding but only weaker interactions. At first, solvent effects on reactivity were studied in terms of some particular solvent parameter. Later on, more sophisticated methods via multiparameter equations were applied such as15 XYZ = XYZ 0 + s(π * +dδ) + aα + bβ + hδ H

[13.1.2]

where XYZ0, s, a, b, and h are solvent-independent coefficients characteristic of the process and indicative of its sensitivity to the accompanying solvent properties. Further, δ is a polarizability correction term equal to 0.0 for nonchlorinated aliphatic solvents, 0.5 for polychlorinated aliphatics, and 1.0 for aromatic solvents. The other parameters have been given above, viz. π*, α, β, and δH are indices of solvent dipolarity/polarizability, Lewis acidity, Lewis basicity, and cavity formation energy, respectively. For the latter, instead of δ H, δH2 should be preferred as suggested from regular solution theory.16 Let us just mention two applications of the linear solvation energy relationship (LSER). The one concerns the solvolysis of tertiary butyl-halides17 log k(ButCl) = -14.60 + 5.10π* + 4.17α + 0.73β + 0.0048δ2H


741

n = 21, r = 0.9973, s = 0.242 and the other deals with the transfer of tetramethylammonium iodide through solvents with methanol as the reference solvent,16 ∆G tr0 = 10.9 − 15.6π∗−6.2α + 0.022δ H n = 18, r = 0.997, s = 0.3 where: n r s

number of solvents correlation coefficient standard deviation

Figure 13.1.3. Relative orbital energy levels for Cu2+ in square planar, tetragonal, and octahedral environments [adapted from ref. 18].

DN = 195.5 - 0.0102ν 0

We will not finish this section without noting that there are also metal complexes available functioning as color indicators of the coordination properties of solvents.18 Thus, Cu(tmen)(acac) ClO4, where tmen = N,N,N’,N’tetramethylethylenediamine and acac = acetylacetonate, can be used as a Lewis-basicity indicator, and Fe(phen)2(CN)2, where phen = 1,10-phenanthroline, as a Lewisacidity indicator. The physical origin of the underlying color changes is sketched in the Figures 13.1.3 and 13.1.4, as modified from ref. 18. These color indicators can be used as a quick method for assessing the coordination properties of solvents, solvent mixtures, and solutes not yet measured. This is very expedient since some classical parameters, particularly the donor numbers, are arduously amenable. The following equation [13.1.3]

n = 12, r = 0.990, s = 1.37 correlates the wave numbers ν 0 (in cm-1) of the visible band of Cu(tmen)(acac)+ and the solvent donor numbers. Similarly, the acceptor numbers are expressed as a function of the wave numbers of the long wavelength absorption of Fe(phen)2(CN)2,

742

Roland Schmid

AN = -133.8 - 0.00933ν 0

[13.1.4]

n = 12, r = 0.980, s = 4.58 13.1.4 DIELECTRIC VS. CHEMICAL APPROACH Although the success of the empirical solvent parameters has tended to downgrade the usefulness of the dielectric approach, there are correlations that have succeeded as exemplified by Figure 13.1.1. It is commonly held that the empirical solvent parameters are superior to dielectric estimates because they are sensitive to short-range phenomena not captured in dielectric measurements. This statement may not be generalized, however, since it depends strongly on the chemical reaction investigated and the choice of solvents. For instance, the rate of the Menschutkin reaction between tripropylamine and methyl Figure 13.1.4. Simplified orbital scheme for the charge transfer tran- iodide in select solvents correlates sition in Fe(phen)2(CN)2 varying with solvation. The diagram, not better with the log ε function than drawn to scale, is adjusted so that π* is constant [adapted from ref. with the solvent acceptor num18]. ber.19 Thus the solution chemists were puzzled for a long time over the question about when and when not the dielectric approach is adequate. In the meantime, this issue has been unraveled, in that dielectric estimates have no relevance to the solvation of positive (partial) charge. Thus, there is no relationship between the free energies of transfer for cations and the dielectric constant.7 Likewise, note the solvent-dependence of the solubilities of sodium chloride (Table 13.1.1) taken from Mayer’s work.19 For instance, the pairs of solvents H2O/PC and DMF/MeCN have similar ε′s but vastly different abilities to dissolve NaCl. In similar terms, the inclusion of a donor number term improves somewhat the correlation in Figure 13.1.1, as may be seen in Figure 13.1.5. This would suggest that the hydrogen of piperidine in the activated complex becomes acidic and is attacked by the strong donor solvents DMF, DMA and DMSO (Scheme 13.1.1).


743

Table 13.1.1. Standard free energies of solution of sodium chloride in various solvents at 25°C. Data of ∆G0solv are from reference 19 ∆G0solv , kJ mol-1

εs

DN

AN

H 2O

-9.0

78.4

18

55

FA

-0.4

109

24

40

NMF

+3.8

182

27

32

MeOH

+14.1

32.6

19

41

DMSO

+14.9

46.7

30

19

DMF

+26.8

36.7

26

16

PC

+44.7

65

15

18

MeCN

+46.8

36

14

19

Solvent

On the other hand, if negative charge is solvated in the absence of positive charge capable of solvation, the dielectric constant is often a pretty good guide to ranking changes in reactivity. As a consequence, the dielectric approach has still its place in organic chemistry while it is doomed to complete failure in inorganic reactions where typically cation solvation is involved. For select solvents, ultimately, the dielectric constant is related to the anion-solvating properties of solvents according to the regression equation4 Figure 13.1.5. Correlation diagram for the same reaction as in Figure 13.1.1 [from ref. 21].

log ε = 0.32 + 0.073 (ANE)

[13.1.5]

n = 31, r = 0.950, s = 0.129 where: ANE

ET-based acceptor numbers, ANE = - 40.52 + 1.29 ET

This equation works also quite well for the aromatics and the halogenated solvents, but it does not hold for the protic solvents. For these, the predicted values of the dielectric constants are orders of magnitude too large, revealing how poorly the associates are dissoci-

744

Roland Schmid

ated by the macroscopically attainable fields. A correlation similar to [13.1.5] has been proposed20 between the gas phase dipole moment and π* µ(D) = 4.3π * −01 .

[13.1.6]

n = 28, r = 0.972, s = 0.3 Along these lines the dielectric and the chemical approach are brought under one roof.4,21 The statement, however, that the terms “good acceptor solvent” and “highly polar solvent” may be used synonymously would seem, though true, to be provocative. 13.1.5 CONCEPTUAL PROBLEMS WITH EMPIRICAL SOLVENT PARAMETERS A highly suspect feature behind the concept of empirical solvent parameters lies in the interpretation of the results in that condensed phase matters are considered from the narrow viewpoint of the solute only with the solvent’s viewpoint notoriously neglected. However, the solute is actually probing the overall action of the solvent, comprising two modes of interactions: solute-solvent (solvation) and solvent-solvent (restructuring) effects of unknown relative contribution. Traditionally, it is held that solvent structure only assumes importance when highly structured solvents, such as water, are involved.22 But this view increasingly turns out to be erroneous. In fact, ignoring solvent-solvent effects, even in aprotic solvents, can lead to wrong conclusions as follows. In the donor-acceptor approach, solutes and solvents are divided into donors and acceptors. Accordingly, correlations found between some property and the solvent donor (acceptor) ability are commonly thought to indicate that positive (negative) charge is involved. In the case of solvent donor effects this statement is actually valid. We are unaware, in fact, of any exception to the rule saying: “Increase in reaction rate with increasing solvent DN implies that positive charge is developed or localized and vice versa”.21 In contrast, correlations with the acceptor number or related scales do not simply point to anion solvation, though this view is commonly held. An example for such type of reasoning concerns the medium effect on the intervalence transition (IT) energy within a certain binuclear, mixed-valence, 5+ cation.23 As the salt effect was found to vary with the solvent AN, anion, that is counterion, solvation in ion pairs was invoked to control the IT energy. A conceptual problem becomes obvious by the at first glance astonishing result that the reduction entropies of essentially non-donor cationic redox couples such as Ru(NH3)63+/2+ are correlated with the solvent AN.24 These authors interpreted this solvent dependence as reflecting changes in solvent-solvent rather than solvent-ligand interactions. That the acceptor number might be related to solvent structure is easy to understand since all solvents of high AN always are good donors (but not vice versa!) and therefore tend to be increasingly self-associated.21 There is since growing evidence that the solvent’s AN and related scales represent ambiguous solvent properties including solvent structural effects instead of measuring anion solvation in an isolated manner. Thus, correlations between Gibbs energies of cation transfer from water to organic solvents and the solvent DN are improved by the inclusion of a term in ET (or a combination of α and π*).25 Consequently Marcus et al. rightly recognized that “ET does not account exclusively for the electron pair acceptance capacity of solvents”.26 In more recent work27 a direct relationship has been


745

found between the solvent reorganizational energy accompanying the excitation of ruthenium(II) cyano complexes and the solvent acceptor number. In the basicity scales, on the other hand, complications by solvent structure are not as obvious. If restriction is to aprotic solvents, as is usual, various scales though obtained under different conditions, are roughly equivalent.21,4 There is for instance a remarkably good relationship between the DN scale (obtained in dilute dichloromethane solution, i.e., with medium effects largely excluded) and the B scale (derived from measurements performed with 0.4 M solutions of MeOD in the various solvents4). The relationship between β and B, on the other hand, separates out into families of solvents.20 Donor measures for protic solvents eventually are hard to assess and often are at considerable variance from one scale to another.28,29 To rationalize the discrepancies, the concept of “bulk donicity” was introduced7 but with little success. Instead, the consideration of structure changes accompanying solvation might better help tackle the problem. Another suspect feature of the common method of interpreting solvent-reactivity correlations is that it is notoriously done in enthalpic (electronic, bond-strength etc.) terms. This way of thinking goes back to the Hughes-Ingold theory. However, many reactions in solution are not controlled by enthalpy changes but instead by entropy. Famous examples are the class of Menschutkin reactions and the solvolysis of t-butyl halides. Both these reaction types are characterized by the development of halide ions in the transition state, which can be considered as ion-pair like. In view of this, rate acceleration observed in good acceptor (or, alternatively, highly polar) solvents seems readily explainable in terms of solvation of the developing halide ion with concomitant carbon-halogen bond weakening. If this is true, most positive activation entropies and highest activation enthalpies should be expected to occur for the poor acceptor solvents. However, a temperature dependence study of the t-butyl halide solvolysis revealed just the opposite.17 This intriguing feature points to changes in solvent structure as a major determinant of the reaction rate with the ionic transition state acting as a structure maker in poor acceptor solvents, and as a structure maker in the protic solvents. It is rather ironic that the expected increase in rate with increasing solvent acceptor strength is a result of the coincidence of two, from the traditional point of view, unorthodox facts: (i) The intrinsic solvation of the developing halide ion disfavors the reaction via the entropy term. However, (ii), the extent of that solvation is greater in the poorly coordinating solvents (providing they are polarizable such as the aromatic solvents and the polyhalogenated hydrocarbons). In keeping with this interpretation, the Menschutkin reaction between benzyl bromide and pyridine is characterized by more negative activation volumina (i.e., stronger contraction of the reacting system in going to the activated complex) in poor acceptor (but polarizable) solvents.30 The importance is evident of studying temperature or pressure dependencies of solvent effects on rate in order to arrive at a physically meaningful interpretation of the correlations. Another problem with the interpretation of multiparameter equations such as [13.1.2] arises since some of the parameters used are not fully independent of one another. As to this, the trend between π* and α has already been mentioned. Similarly, the δH parameter displays some connection to the polarity indices.31,32 Virtually, the various parameters feature just different blends of more fundamental intermolecular forces (see below). Because of this, the interpretations of empirical solvent-reactivity correlations are often based more on intuition or preconceived opinion than on physically defined interaction mechanisms. As it

746

Roland Schmid

turns out, polar solvation has traditionally been overemphasized relative to nonpolar solvation (dispersion and induction), which is appreciable even in polar solvents. The conceptual problems of the empirical solvent parameters summarized: • The solvent acceptor number and other “polarity” scales include appreciable, perhaps predominant, contributions from solvent structure changes rather than merely measuring anion solvation. • Care is urged in a rash interpretation of solvent-reactivity correlations in enthalpic terms, instead of entropic, before temperature-dependence data are available. Actually, free energy alone masks the underlying physics and fails to provide predictive power for more complex situations. • Unfortunately, the parameters used in LSER’s sometimes tend to be roughly related to one another, featuring just different blends of more fundamental intermolecular forces. Not seldom, fortuitous cancellations make molecular behavior in liquids seemingly simple (see below). Further progress would be gained if the various interaction modes could be separated by means of molecular models. This scheme is in fact taking shape in current years giving rise to a new era of tackling solvent effects as follows. 13.1.6 THE PHYSICAL APPROACH There was a saying that the nineteenth century was the era of the gaseous state, the twentieth century of the solid state, and that perhaps by the twenty-first century we may understand something about liquids.33 Fortunately, this view is unduly pessimistic, since theories of the liquid state have actively been making breath-taking progress. In the meantime, not only equations of state of simple liquids, that is in the absence of specific solvent-solvent interactions,34-36 but also calculations of simple forms of intermolecular interactions are becoming available. On this basis, a novel approach to treating solvent effects is emerging, which we may call the physical approach. This way of description is capable of significantly changing the traditionally accepted methods of research in chemistry and ultimately will lay the foundations of the understanding of chemical events from first principles. A guiding principle of these theories is recognition of the importance of packing effects in liquids. It is now well-established that short-ranged repulsive forces implicit in the packing of hard objects, such as spheres or dumbbells, largely determine the structural and dynamic properties of liquids.37 It may be noted in this context that the roots of the idea of repulsive forces reach back to Newton who argued that an elastic fluid must be constituted of small particles or atoms of matter, which repel each other by a force increasing in proportion as their distance diminishes. Since this idea stimulated Dalton, we can say that the very existence of liquids helped to pave the way for formulating modern atomic theory with Newton granting the position of its “grandfather”.38 Since the venerable view of van der Waals, an intermolecular potential composed of repulsive and attractive contributions is a fundamental ingredient of modern theories of the liquid state. While the attractive interaction potential is not precisely known, the repulsive part, because of changing sharply with distance, is treatable by a common formalism in terms of the packing density η, that is the fraction of space occupied by the liquid molecules. The packing fraction is a key parameter in liquid state theories and is in turn related in a simple way to the hard sphere (HS) diameter σ in a spherical representation of the molecules comprising the fluid:


η = πρσ 3 / 6 = ρVHS

747

[13.1.7]

where: η ρ σ VHS

packing density number density N/V= number of particles per unit volume HS diameter HS volume

For the determination of σ (and hence η), the most direct method is arguably that based on inert gas solubility data.39,40 However, in view of the arduousness involved and the uncertainties in both the extrapolation procedure and the experimental solubilities, it is natural to look out for alternatives. From the various suggestions,41,42 a convenient way is to adjust σ such that the computed value of some selected thermodynamic quantity, related to σ, is consistent with experiment. The hitherto likely best method43 is the following: To diminish effects of attraction, the property chosen should probe primarily repulsive forces rather than attractions. Since the low compressibility of the condensed phase is due to short-range repulsive forces, the isothermal compressibility βT = -(1/V)(∂V/∂P)T might be a suitable candidate, in the framework of the generalized van der Waals (vdW) equation of state βT (RT / V)Q r = 1

[13.1.8]

where Qr is the density derivative of the compressibility factor of a suitable reference system. In the work referred to, the reference system adopted is that of polar-polarizable spheres in a mean field,   5η 2 − 2η 3 Q r = 2 − 1 − 2Z µ  4   (1 − η)

[13.1.9]

where Zµ = compressibility factor due to dipole-dipole forces,43 which is important only for a few solvents such as MeCN and MeNO2. The HS diameters so determined are found to be in excellent agreement with those derived from inert gas solubilities. It may be noted that the method of Ben-Amotz and Willis,44 also based on βT, uses the nonpolar HS liquid as the reference and, therefore, is applicable only to liquids of weak dipole-dipole forces. Of course, as the reference potential approaches that of the real liquid, the HS diameter of the reference liquid should more closely approximate the actual hard-core length. Finally, because of its popularity, an older method should be mentioned that relies on the isobaric expansibility αp as the probe, but this method is inadequate for polar liquids. It turns out that solvent expansibility is appreciably determined by attractions. Some values of η and σ are shown in Table 13.1.2 including the two extreme cases. Actually, water and n-hexadecane have the lowest and highest packing density, respectively, of the common solvents. As is seen, there is an appreciable free volume, which may be expressed by the volume fraction η − η 0 , where η 0 is the maximum value of η calculated for the face-centered cubic packing of HS molecules where all molecules are in contact with each other is η 0 = π 2 / 6 = 0.74. Thus, 1 - η 0 corresponds to the minimum of unoccupied volume. Since η typically is around 0.5, about a quarter of the total liquid volume is empty enabling solvent molecules to change their coordinates and hence local density fluctuations to occur.

748

Roland Schmid

Packing density η minimum ≈0 perfect gas

0.4 - 0.6 liquids

maximum 0.74 cubic close packed

Table 13.1.2. Packing densities in some liquids

These considerations ultimately offer the basis of a genuinely Liquid free volume, % molecular theory of solvent effects, η as compared to a mean-field theory. 0.41 59 H2O Thus, packing and repacking effects 0.50 50 n-C6 accompanying chemical reactions have to be taken into account for any Benzene 0.51 49 realistic view of the solvent’s role MeOH 0.41 59 played in chemical reactions to be at0.47 53 Et2O tained. The well-known cavity formation energy is the work done 0.62 38 n-C16 against intermolecular repulsions. At present, this energy is calculated for spherical cavities by the Boublik-Mansoori-Carnahan-Starling-Leland (BMCSL) mixed HS equation of state45,46 ∆G rep RT

=2

ηd ( −d 2 + d + 1) ηd 3 ηd 2 +3 +3 + ( −2d 3 + 3d 2 − 1) ln( 1− η) 3 2 ( 1− η) ( 1− η) ( 1− η)

[13.1.10]

where d = σ 0 / σ is the relative solute size (σ0 is the solute HS diameter, and σ is the solvent diameter). Quite recently, a modification of this equation has been suggested for high liquid densities and large solute sizes.47 Notice that under isochoric conditions the free energy of cavity formation is a totally entropic quantity. Ravi et al48 have carried out an analysis of a model dissociation reaction (Br2 → 2Br) dissolved in a Lennard-Jones solvent (Ne, Ar, and Xe). That and the previous work49 demonstrated that solvent structure contributes significantly to both chemical reaction volumes (which are defined as the pressure derivatives of reaction free energies) and free energies, even in systems containing no electrostatic or dispersion long-ranged solvent-solute interactions. Let us now turn to the more difficult case of intermolecular attractive forces. These may be subdivided into: Long-ranged or unspecific • dispersion • induction • dipole-dipole • higher multipole Short-ranged or specific • electron overlap (charge transfer) • H-bonding For the first three ones (dispersion, induction, dipole-dipole forces) adequate calculations are just around the corner. Let us give some definitions.


749

Dispersion forces are the result of the dipolar interactions between the virtually excited dipole moments of the solute and the solvent, resulting in a nonzero molecular polarizability. Although the average of every induced dipole is zero, the average of the product of two induced dipoles is nonzero (Figure 13.1.6). Induction forces are caused by the interaction of the permanent solvent dipole with the solvent dipoles induced by the solute and solvent field (Figure 13.1.7). Sometimes it is stated that dispersion is a quantum mechanical effect and induction is not. Thus, Figure 13.1.6 some clarifying comments are at place here. From the general viewpoint, all effects including polarizability are quantum mechanical in their origin because the polarizability of atoms and molecules is a quantum mechanical quantity and can be assessed only in the framework of quantum mechanics. However, once calculated, one can think of polarizability in classical terms representing a quantum molecular object as a classical oscillator with the mass equal to the polarizability, which is not specified in the classical framework. This is definitely wrong from a fundamental viewpoint, but, as it usually appears with harmonic models, a quantum mechanical calculation and such a primitive classical model give basically the same results about the induction Figure 13.1.7 matter. Now, if we implement this classical model, we would easily come up with the induction potential. However, the dispersion interaction will be absent. The point is that to get dispersions, one needs to switch back to the quantum mechanical description where both inductions and dispersions naturally appear. Thus the quantum oscillator may be used resulting in both types of potentials.50 If in the same procedure one switches to the classical limit (which is equivalent to putting the Plank constant zero) one would get only inductions. The calculation of the dispersive solvation energy is based on perturbation theories following the Chandler-Andersen-Weeks51 or Barker-Henderson52 formalisms, in which long-range attractive interactions are treated as perturbations to the properties of a hard body reference system. Essentially, perturbative theories of fluids are a modern version of van der Waals theory.53 In the papers reviewed here, the Barker-Henderson approach was utilized with the following input parameters: Lennard-Jones (LJ) energies for the solvent, for which reliable values are now available, the HS diameters of solvent and solute, the solvent polarizability, and the ionization potentials of solute and solvent. A weak point is that in order to get the solute-solvent LJ parameters from the solute and solvent components, some combining rule has to be utilized. However, the commonly applied combining rules appear to be adequate only if solute and solvent molecules are similar in size. For the case of particles appreciably different both in LJ energy and size, the suggestion has been made to use an empirical scaling by introducing empirical coefficients so as to obtain agreement be-

750

Roland Schmid

tween calculated and experimental solvation energies for selected inert gases and nonpolar large solutes.54 In the paper referred to,54 the relevance of the theoretical considerations has been tested on experimental solvation free energies of nitromethane as the solute in select solvents. The total solvation energy is a competition of the positive cavity formation energy and the negative solvation energy of dispersion and dipolar forces, ∆G = ∆G cav + ∆G disp + ∆G dipolar

[13.1.11]

where the dipolar term includes permanent and induced dipole interactions. The nitromethane molecule is represented by the parameters of the HS diameter σ = 4.36 Å, the gas-phase dipole moment µ = 3.57 D, the polarizability α = 4.95 Å , and the LJ energy εLJ/k=391K. Further, the solvent is modeled by spherical hard molecules of spherical polarizability, centered dipole moment, and central dispersion potential. To calculate the dipolar response, the Padé approximation was applied for the chemical potential of solvation in the dipolar liquid and then extended to a polarizable fluid according to the procedure of Wertheim. The basic idea of the Wertheim theory is to replace the polarizable liquid of coupled induced dipoles with a fictitious fluid with an effective dipole moment calculated in a self-consistent manner. Further, the Padé form is a simple analytical way to describe the dependence of the dipolar response on solvent polarity, solvent density, and solute/solvent size ratio. The theory/experiment agreement of the net solvation free energy is acceptable as seen in Table 13.1.3 where solvent ordering is according to the dielectric constant. Note that the contribution of dispersion forces is considerable even in strongly polar solvents. Table 13.1.3. Thermodynamic potentials (kJ/mol) of dissolution of nitromethane at 25°C. Data are from reference 54 εs

∆Gcav

∆Gdisp

∆Gdipolar

∆G(calc)

∆G(exp)

n-C6

1.9

23.0

-32.9

-2.2

-12.1

-12.1

c-C6

2.0

28.1

-38.1

-2.8

-12.7

-12.0

Et3N

2.4

24.2

-33.5

-2.9

-12.2

-15.2

Et2O

4.2

22.6

-33.4

-5.8

-16.5

-17.5

EtOAc

6.0

28.0

-38.5

-9.7

-20.1

-21.2

THF

7.5

32.5

-41.5

-12.2

-21.1

-21.3

c-hexanone

15.5

35.1

-39.8

-17.9

-22.6

-21.8

2-butanone

17.9

28.3

-33.9

-18.9

-24.5

-21.9

Acetone

20.7

27.8

-31.2

-22.1

-25.5

-22.5

DMF

36.7

38.1

-32.1

-28.5

-22.5

-23.7

DMSO

46.7

41.9

-31.1

-31.0

-20.2

-23.6

Solvent

With an adequate treating of simple forms of intermolecular attractions becoming available, there is currently great interest to making a connection between the empirical scales and solvation theory. Of course, the large, and reliable, experimental databases on


751

empirical parameters are highly attractive for theoreticians for testing their computational models and improving their predictive power. At present, the solvatochromic scales are under considerable scrutiny. Thus, in a recent thermodynamic analysis, Matyushov et al.55 analyzed the two very popular polarity scales, ET(30) and π*, based on the solvent-induced shift of electronic absorption transitions (Scheme 13.1.2) Solvatochromism has its origin in changes in both dipole moment and polarizability of the dye upon electronic excitation provoking differential solvation of the ground and excited states. The dipole moment, µe, of the excited state can be either smaller or larger than the ground state value µg. In the former case one speaks about a negatively solvatochromic dye such as betaine-30, whereas Table 13.1.4. Dye properties used in the the π* dye 4-nitroanisole is positively solvatochromic. Thus, polar calculations. Data are from reference 55 solvent molecules produce a red Molecular parameter Betaine-30 4-Nitroanisole shift (lower energy) in the former and a blue shift (higher energy) in Vacuum energy gap (eV) 1.62 4.49 the latter. On the other hand, 6.4 4.5 R0 (Å) polarizability arguably always increases upon excitation. Disper68 15 α g (Å) sion interactions, therefore, would 61 6 ∆α (Å) produce a red shift proportional to 14.8 4.7 mg (D) ∆α = αe - αg of the dye. In other words, the excited state is stabi6.2 12.9 me (D) lized through strengthening of -8.6 +8.2 ∆µ (D) dispersive coupling. Finally, the relative contributions of dispersion and dipolar interactions will depend on the size of the dye molecules with dispersive forces becoming increasingly important the larger the solute. Along these lines, the dye properties entering the calculations are given in Table 13.1.4. The purpose of the analysis was to determine how well the description in terms of “trivial” dipolar and dispersion forces can reproduce the solvent dependence of the absorption energies (and thereby, by difference to experiment, expose the magnitude of specific forces), [13.1.12] £ωabs = ∆ + ∆E rep + ∆E disp + ∆E dipolar + ∆E ss where

£ω abs ∆ ∆Erep ∆Edisp ∆Edipolar ∆Ess

absorption energy vacuum energy gap shift due to repulsion solute-solvent interactions (taken to be zero) shift due to dispersion interactions shift due to dipolar forces of permanent and induced dipoles solvent reorganization energy

For the detailed and arduous calculation procedure, the reader may consult the paper cited. Here, let us just make a few general comments. The solvent influence on intramolecular optical excitation is treated by implementing the perturbation expansion

752

Roland Schmid

over the solute-solvent attractions. The reference system for the perturbation expansion is chosen to be the HS liquid with the imbedded hard core of the solute. It should be noted for clarity that ∆Edisp and ∆Edipolar are additive due to different symmetries: dispersion force is non-directed (i.e., is a scalar quantity), and dipolar force is directed (i.e., is a vector). In other words, the attractive intermolecular potential can be split into a radial and an angle-dependent part. In modeling the solvent action on the optical excitation, the solute-solvent interactions have to be dissected into electronic (inertialess) (dispersion, induction, charge-transfer) and molecular (inertial) (molecular orientations, molecular packing) modes. The idea is that the inertial modes are frozen on the time scale of the electronic transition. This is the Franck-Condon principle with such types of transitions called vertical transitions. Thus the excited solute is to be considered as a Frank-Condon state, which is equilibrated only to the electronic modes, whereas the inertial modes remain equilibrated to the ground state. According to the frozen solvent configuration, the dipolar contribution is represented as the sum of two terms corresponding to the two separate time scales of the solvent, (i) the variation in the solvation potential due to the fast electronic degrees of freedom, and (ii) the work needed to change the solute permanent dipole moment to the excited state value in a frozen solvent field. The latter is calculated for accommodating the solute ground state in the solvent given by orientations and local packing of the permanent solvent dipoles. Finally, the solvent reorganization energy, which is the difference of the average solvent-solvent interaction energy in going from the ground state to the excited state, is extracted by treating the variation with temperature of the absorption energy. Unfortunately, experimental thermochromic coefficients are available for a few solvents only. The following results of the calculations are relevant. While the contributions of dispersions and inductions are comparable in the π* scale, inductions are overshadowed in the ET(30) values. Both effects reinforce each other in π*, producing the well-known red shift. For the ET(30) scale, the effects due to dispersion and dipolar solvation have opposite signs making the red shift for nonpolar solvents switch to the blue for polar solvents. Furthermore, there is overall reasonable agreement between theory and experiment for both dyes, as far as the nonpolar and select solvents are concerned, but there are also discrepant solvent classes pointing to other kinds of solute-solvent interactions not accounted for in the model. Thus, the predicted ET(30) values for protic solvents are uniformly too low, revealing a decrease in H-bonding interactions of the excited state with lowered dipole moment. Another intriguing observation is that the calculated π* values of the aromatic and chlorinated solvents are throughout too high (in contrast to the ET(30) case). Clearly, these deviations, reminiscent of the shape of the plots such as Figure 13.1.2, may not be explained in terms of polarizability as traditionally done (see above), since this solvent property has been adequately accommodated in the present model via the induction potential. Instead, the theory/experiment discord may be rationalized in either of two ways. One reason for the additional solvating force can be sought in terms of solute-solvent π overlap resulting in exciplex formation. Charge-transfer (CT) interactions are increased between the solvent and the more delocalized excited state.55 The alternative, and arguably more reasonable, view considers the quadrupole moment which Figure 13.1.8


753

is substantial for both solvent classes.56 Recently, this latter explanation in terms of dipole-quadrupole interactions is favored (Fig 13.1.8). It is well-known that the interaction energy falls off more rapidly the higher the order of the multipole. Thus, for the interaction of an n-pole with an m-pole, the potential energy varies with distance as E ∝ 1/(rn+m+1). The reason for the faster decrease is that the array of charges seems to blend into neutrality more rapidly with distance the higher the number of individual charges contributing to the multipole. Consequently, quadrupolar forces die off faster than dipolar forces. It has been calculated that small solute dipoles are even more effectively solvated by solvent quadrupoles than by solvent dipoles.57 In these terms it is understandable that quadrupolar contributions are more important in the π* than in the ET(30) scale. Similarly, triethylphosphine oxide, the probe solute of the acceptor number scale, is much smaller than betaine(30) and thus might be more sensitive to quadrupolar solvation. Thus, at long last, the shape of Figure 13.1.2 and similar ones seems rationalized. Note by the way that the quadrupole and CT mechanisms reflect, respectively, inertial and inertialess solvation pathways, and hence could be distinguished by a comparative analysis of absorption and fluorescence shifts (Stokes shift analysis). However, for 4-nitroanisole fluorescence data are not available. Reverting once more to the thermodynamic analysis of the π* and ET(30) scales referred to above, it should be mentioned that there are also other theoretical treatments of the solvatochromism of betaine(30). Actually, in a very recent computer simulation,58 the large polarizability change ∆α (nearly 2-fold, see Table 13.1.4) upon the excitation of betaine(30) has been (correctly) questioned. (According to a rule of thumb, the increase in polarizability upon excitation is proportional to the ground state polarizability, on the order ∆α ≈ 0.25α g .50) Unfortunately, Matyushov et al.55 derived this high value of ∆α = 61 Å3 from an analysis of experimental absorption energies based on aromatic, instead of alkane, solvents as nonpolar reference solvents. A lower value of ∆α would diminish the importance of dispersion interactions. Further theoretical and computational studies of betaine(30) of the ET(30) scale are reviewed by Mente and Maroncelli.58 Despite several differences in opinion obvious in these papers, an adequate treatment of at least the nonspecific components of solvatochromism would seem to be “just around the corner”. Finally, a suggestion should be mentioned on using the calculated π* values taken from ref. 55 as a descriptor of nonspecific solvent effects.59 However, this is not meaningful since these values are just a particular blend of inductive, dispersive, and dipole-dipole forces. 13.1.7 SOME HIGHLIGHTS OF RECENT INVESTIGATIONS The like dissolves like rule The buzzword “polarity”, derived from the dielectric approach, is certainly the most popular word dealing with solvent effects. It is the basis for the famous rule of thumb “similia similibus solvuntur” (“like dissolves like”) applied for discussing solubility and miscibility. Unfortunately, this rule has many exceptions. For instance, methanol and toluene, with dielectric constants of 32.6 and 2.4, respectively, are miscible, as are water (78.4) and isopropanol (18.3). The problem lies in exactly what is meant by a “like” solvent. Originally, the term “polarity” was meant to be an abbreviation of “static dipolarity” and was thus associated with solely the dielectric properties of the solvent. Later on, with the advent

754

Roland Schmid

of the empirical solvent parameters, it has assumed a broader meaning, sometimes even that of the overall solvating power.13 With this definition, however, the term “polarity” is virtually superfluous. Clearly, neither the dielectric constant nor the dipole moment is an adequate means to define polarity. The reason is that there are liquids whose constituent molecules have no net dipole moment, for symmetry reasons, but nevertheless have local polar bonds. This class of solvents, already mentioned above, comprises just the notorious troublemakers in solvent reactivity correlations, namely the aromatic and chlorinated solvents. These solvents, called “nondipolar” in the literature,60 stabilize charge due to higher solvent multipoles (in addition to dispersive forces) like benzene (“quadrupolar”) and carbon tetrachloride (“octupolar”). Of this class, the quadrupolar solvents are of primary importance. Thus, the gas-phase binding energy between of K+ and benzene is even slightly greater than that of K+-water. The interaction between the cation and the benzene molecule is primarily electrostatic in nature, with the ion-quadrupole interaction accounting for 60% of the binding energy.61 This effect is size-dependent: Whereas at K+ benzene will displace some water molecules from direct contact with the ion, Na +aq is resistant towards dehydration in an aromatic environment, giving rise to selectivity in some K+ channel proteins.62 For the polarity of the C-H bonds, it should be remembered that electronegativity is not an intrinsic property of an atom, but instead varies with hybridization. Only the C(sp3)-H bond can be considered as truly nonpolar, but not so the C(sp2)-H bond.63 Finally, ethine has hydrogen atoms that are definitely acidic. It should further be mentioned that higher moments or local polarities cannot produce a macroscopic polarization and thus be detected in infinite wavelength dielectric experiments yielding a static dielectric constant close to the squared refractive index. Because of their short range, quadrupolar interactions do not directly contribute to the dielectric constant, but are reflected only in the Kirkwood gK factor that decreases due to breaking the angular dipole-dipole correlations with increasing quadrupolar strength. In these terms it is strongly recommended to redefine the term polarity. Instead of meaning solely dipolarity, it should also include higher multipolar properties, polarity = dipolarity + quadrupolarity + octupolarity

[13.1.13]

This appears to be a better scheme than distinguishing between truly nonpolar and nondipolar solvents.56 A polar molecule can be defined as having a strongly polar bond, but need not necessarily be a dipole. In this framework, the solvating power of the “nondipolar” solvents need no longer be viewed as anomalous or as essentially dependent on specific solvation effects.10 Beyond this it should be emphasized that many liquids have both a dipole moment and a quadrupole moment, water for example. However, for dipolar solvents such as acetonitrile, acetone, and dimethyl sulfoxide, the dipolar solvation mechanism will be prevailing. For less dipolar solvents, like tetrahydrofuran, quadrupoles and dipoles might equally contribute to the solvation energetics.57 Notwithstanding this modified definition, the problem with polarity remains in that positive and negative charge solvation is not distinguished. As already pointed out above, there is no general relationship between polarity and the cation solvation tendency. For example, although nitromethane (MeNO2) and DMF have the same dielectric constant, the extent of ion pairing in MeNO2 is much greater than that in DMF. This observation is


755

attributed to the weak basicity of MeNO2 which poorly solvates cations. As a result, ion pairing is stronger in MeNO2 in spite of the fact that long-range ion-ion interactions in the two solvents are equal. Finally, a potential problem with polarity rests in the fact that this term is typically associated with enthalpy. But caution is urged in interpreting the like-dissolves-like rule in terms of enthalpy. It is often stated for example that nonpolar liquids such as octane and carbon tetrachloride are miscible because the molecules are held together by weak dispersion forces. However, spontaneous mixing of the two phases is driven not by enthalpy, but by entropy. Water’s anomalies The outstanding properties and anomalies of water have fascinated and likewise intrigued physicists and physical chemists for a long time. During the past decades much effort has been devoted to finding phenomenological models that explain the (roughly ten) anomalous thermodynamic and kinetic properties, including the density maximum at 4°C, the expansion upon freezing, the isothermal compressibility minimum at 46°C, the high heat capacity, the decrease of viscosity with pressure, and the remarkable variety of crystalline structures. Furthermore, isotope effects on the densities and transport properties do not possess the ordinary mass or square-root-mass behavior. Some of these properties are known from long ago, but their origin has been controversial. From the increasingly unmanageable number of papers that have been published on the topic, let us quote only a few that appear to be essential. Above all, it seems to be clear that the exceptional behavior of water is not simply due to hydrogen bonding, but instead due to additional “trivial” vdW forces as present in any liquid. A hydrogen bond occurs when a hydrogen atom is shared between generally two electronegative atoms; vdW attractions arise from interactions among fixed or induced dipoles. The superimposition and competition of both is satisfactorily accommodated in the framework of a “mixture model”. The mixture model for liquid water, promoted in an embryonic form by Röntgen64 over a century ago, but later discredited by Kauzmann65 and others,66 is increasingly gaining ground. Accordingly there are supposed to be two major types of intermolecular bonding configurations, an open bonding form, with a low density, such as occurs in ice-Ih, plus a dense bonding form, such as occurs in the most thermodynamically stable dense forms of ice, e.g., ice-II, -III, -V, and -VI.67 In these terms, water has many properties of the glassy states associated with multiple hydrogen-bond network structures.68 Clearly, for fluid properties, discrete units, (H2O)n, which can move independently of each other are required. The clusters could well be octamers dissociating into tetramers, or decamers dissociating into pentamers.69,70 (Note by the way that the unit cell of ice contains eight water molecules.) However, this mixture is not conceived to be a mixture of ices, but rather is a dynamic (rapidly fluctuating) mixture of intermolecular bonding types found in the polymorphs of ice. A theoretical study of the dynamics of liquid water has shown that there exist local collective motions of water molecules and fluctuation associated with hydrogen bond rearrangement dynamics.68 The half-life of a single H bond estimated from transition theory is about 2x10-10 s at 300K.71 In view of this tiny lifetime it seems more relevant to identify the two mixtures not in terms of different cluster sizes, but rather in terms of two different bonding modes. Thus, there is a competition between dispersion interactions that favor random dense states and hydrogen bonding that favors ordered open states. Experimental verification of the two types of bonding has been reviewed by Cho et al.72

756

Roland Schmid

From X-ray and neutron scattering, the open structure is characterized by an inner tetrahedral cage of four water molecules surrounding a central molecule, with the nearest-neighbor OLO distance of about 2.8 Å. This distance, as well as the nearest-neighbor count of four, remains essentially intact in both in all crystalline ice polymorphs and in the liquid water up to near the boiling point. In this open tetrahedral network the second-neighbor OLO distance is found at 4.5 Å. However, and most intriguing, another peak in the OLO radial distribution function (RDF), derived from a nonstandard structural approach (ITD), is found near 3.4 Å,73 signaling a more compact packing than in an ordinary H-bonding structure. This dense bonding form is affected through dispersive OLO interactions supplanting H-bonding. Note, however, that in this array the H bonds may not be envisaged as being really broken but instead as being only bent. This claim is substantiated by a sophisticated analysis of vibrational Raman spectra74 and mid-IR spectra70 pointing to the existence of essentially two types of H bonds differing in strength, with bent H bonds being weaker than normal (i.e., linear) H bonds. It should be mentioned that the 3.4 Å feature is hidden by the ordinary minimum of open tetrahedral contributions to the RDF. Because of this, the ordinary integration procedure yields coordination numbers greater than four,75 which confuses the actual situation. Instead, it is the outer structure that is changing whereas the inner coordination sphere remains largely invariant. Even liquid water has much of the tetrahedral H-bonding network of ice I. As temperature, or pressure, is raised, the open tetrahedral hydrogen bonding structure becomes relatively less stable and begins to break down, creating more of the dense structure. Actually all the anomalous properties of water can be rationalized on the basis of this open → dense transformation. An extremum occurs if two opposing effects are superimposed. The density maximum, for instance, arises from the increase in density due to the thermal open → dense transformation and the decrease in density due to a normal thermal expansion.72 As early as 1978 Benson postulated that the abnormal heat capacity of water is due to an isomerization reaction.76 A clear explanation of the density anomaly is given by Silverstein et al.77: “The relatively low density of ice is due to the fact that H-bonding is stronger than the vdW interactions. Optimal H-bonding is incommensurate with the tighter packing that would be favored by vdW interactions. Ice melts when the thermal energy is sufficient to disrupt and disorder the H-bonds, broadening the distribution of H-bond angles and lengths. Now among this broadened H-bond distribution, the vdW interactions favor those conformations of the system that have higher density. Hence liquid water is denser than ice. Heating liquid water continues to further deform hydrogen bonds and increase the density up to the density anomaly temperature. Further increase of temperature beyond the density anomaly weakens both H bonds and vdW bonds, thus reducing the density, as in simpler liquids.” The same authors commented on the high heat capacity of water as follows: “Since the heat capacity is defined as CP = (∂H/∂T)P the heat capacity describes the extent to which some kind of bonds are broken (increasing enthalpy) with increasing temperature. Breaking bonds is an energy storage mechanism. The heat capacity is low in the ice phase because thermal energy at those temperatures is too small to disrupt the H bonds. The heat capacity peaks at the melting temperature where the solid-like H bonds of ice are weakened to become the liquid-like H bonds of liquid water. The reason liquid water has a higher heat capacity than vdW liquids have is because water has an additional energy storage mechanism, namely the H bonds, that can also be disrupted by thermal energies.”


757

Summed up, it appears that any concept to be used in a realistic study of water should have as a fundamental ingredient the competition between expanded, less dense structures, and compressed, more dense ones. Thus, the outer structure in the total potential of water should be characterized by a double minimum: open tetrahedral structure with a second-neighbor OLO distance of 4.5 Å and a bent H-bond structure with an OLO non-H-bonded distance of about 3.4 Å. Actually, according to a quite recent theoretical study, all of the anomalous properties of water are qualitatively explainable by the existence of two competing equilibrium values for the interparticle distance.78 Along these lines the traditional point of view as to the structure of water is dramatically upset. Beyond that, also the classical description of the hydrogen bond needs revision. In contrast to a purely electrostatic bonding, quite recent Compton X-ray scattering studies have demonstrated that the hydrogen bonds in ice have substantial covalent character,79 as already suggested by Pauling in the 1930s.80 In overall terms, a hydrogen bond is comprised of electrostatic, dispersion, charge-transfer, and steric repulsion interactions. Similarly, there are charge-transfer interactions between biological complexes and water81 that could have a significant impact on the understanding of biomolecules in aqueous solution. Finally, we return to the physical meaning of the large difference, for the protic solvents, between the cohesive energy density εc and the internal pressure Pi, quoted in section 13.1.3. For water this difference is highest with the factor εc/Pi equal to 15.3. At first glance this would seem explainable in the framework of the mixture model if H bonding is insensitive to a small volume expansion. However, one should have in mind the whole pattern of the relationship between the two quantities. Thus, εc - Pi is negative for nonpolar liquids, relatively small (positive or negative) for polar non-associated liquids, and strongly positive for H-bonded liquids. A more rigorous treatment41 using the relations, Pi = (∂U/∂V)T = T(∂P/ ∂T)V - P and the thermodynamic identity (∂S/∂V)T = (∂P/∂T)V reveals that the relationship is not as simple and may be represented by the following equation with dispersion detached from the other types of association, ε c − Pi = P −

Udisp  RT  2 Z 0 +  − ρUass + ρ T (∂S ass / ∂ρ)T V  RT 

[13.1.14]

where: P V Z0 Udisp Uass ρ Sass

external pressure liquid volume compressibility factor due to intermolecular repulsion potential of dispersion potential of association excluding dispersion liquid number density entropy of association excluding dispersion

With the aid of this equation we readily understand the different ranges of εc - Pi found for the different solvent classes. Thus, for the nonpolar liquids, the last two terms are negligible, and for the usual values, Zo ≈ 10, -Udisp/RT ≈ 8, and V ≈ 150 cm3, we obtain the typical order of εc - Pi ≈ -(300 - 400) atm (equal to -(30 - 40) J cm-3, since 1 J cm-3 ≡ 9.875 atm). For moderately polar liquids, only the last term remains small, while the internal energy of dipolar forces is already appreciable -ρUpolar ≈ (200-500) atm giving the usual magnitude of εc - Pi. For H-bonded liquids, ultimately, the last term turns out to dominate reflecting the large increase in entropy of a net of H-bonds upon a small decrease in liquid density.

758

Roland Schmid

The hydrophobic effect In some respects the hydrophobic effect may be considered as the converse of the like-dissolves-like rule. The term hydrophobic effect refers to the unusual behavior of water towards nonpolar solutes. Unlike simple organic solvents, the insertion of nonpolar solutes into water is (1) strongly unfavorWater CCl4 able though slightly favored by enthalpy, but (2) - 64.4 - 7.1 ∆S*, J/mol K strongly opposed by a large, negative change in entropy at room temperature, and (3) accompanied by 300∆S*, kJ/mol - 19.3 - 2.1 a large positive heat capacity. An example is given - 10.9 - 1.2 ∆H°, kJ/mol in Table 13.1.5 for the thermodynamic properties of + 8.4 + 0.9 ∆G*, kJ/mol methane dissolved in water and in carbon tetrachloride. In dealing with the entropy (and free energy) of 217.5 0 to 42 ∆Cp, J/mol K hydration, a brief remark on the choice of standard states is in order. The standard molar entropy of dissolution, ∆solvS° pertains to the transfer from a 1 atm gas state to a 1 mol L-1 solution and hence includes compression of the gas phase from 1 mol contained in 24.61 L (at 300 K) to 1 mol present in 1 L. Since theoretical calculations disregard volume contributions, it is proper to exclude the entropy of compression equal to -Rln24.61 = -26.63 J K-1 mol-1, and instead to deal with ∆solvS*.82 Thus, Table 13.1.5. Solution thermodynamics of methane in water and carbon tetrachloride at 25°C. [Data from T. Lazaridis and M. E. Paulaitis, J. Phys. Chem., 96, 3847 (1992) and ref. 108]

∆solv S * = ∆solv S o + 26.63 JK −1mol −1

[13.1.15]

∆solv G *300 = ∆solv H o − 300∆solv S *

[13.1.16]

and

Hydrophobicity forms the basis for many important chemical phenomena including the cleaning action of soaps and detergents, the influence of surfactants on surface tension, the immiscibility of nonpolar substances in water,83 the formation of biological membranes and micelles,84,85 the folding of biological macromolecules in water,86 clathrate hydrate formation,87 and the binding of a drug to its receptor.88 Of these, particularly intriguing is the stabilization of protein structure due to the hydrophobicity of nonpolar groups. Hydrophobic interactions are considerably involved in self-assembly, leading to the aggregation of nonpolar solutes, or equivalently, to the tendency of nonpolar oligomers to adopt chain conformations in water relative to a nonpolar solvent.89 Ever-increasing theoretical work within the last years is being lifting the veil of secrecy about the molecular details of the hydrophobic effect, a subject of vigorous debate. Specifically, the scientific community would eagerly like to decide whether the loss in entropy stems from the water-water or the water-solute correlations. There are two concepts. The older one is the clathrate cage model reaching back to the “iceberg” hypothesis of Frank and Evans,90 and the other, newer one, is the cavity-based model. It should be stressed here that the vast literature on the topic is virtually impossible to survey comprehensively. In the following we will cite only a few papers (and references therein) that paved the way to the present state of the art.


759

The clathrate cage model states that the structure of water is strengthened around a hydrophobic solute, thus causing a large unfavorable entropic effect. The surrounding water molecules adopt only a few orientations (low entropy) to avoid “wasting” hydrogen bonds, with all water configurations fully H-bonded (low energy). There is experimental evidence of structure strengthening, such as NMR and FT-IR studies,91 NMR relaxation,92 dielectric relaxation,93 and HPLC.94 A very common conclusion is that the small solubility of nonpolar solutes in water is due to this structuring process. In the cavity-based model the hard core of water molecules is more important to the hydrophobic effect than H-bonding of water. The process of solvation is dissected into two components, the formation of a cavity in the water to accommodate the solute and the interaction of the solute with the water molecules. The creation of a cavity reduces the volume of the translational motion of the solvent particles. This causes an unfavorable entropic effect. The total entropy of cavity formation at constant pressure54 ∆S cav, P = ρα p (∂∆G cav / ∂ρ)T − ∆G cav / T

[13.1.17]

where ∆Scav,P ρ ∆Gcav

cavity formation entropy at constant pressure liquid number density free energy of cavity formation

is the result of the opposing nature of the (positive) liquid expansibility term and the (negative) chemical potential summand. Along these lines the large and negative entropy of cavity formation in water is traced to two particular properties of water: the small molecular size (σ = 2.87 Å) and the low expansibility (αp = 0.26x10-3 K-1), with the latter having the greater impact. It is interesting to note that in both aspects water is extraordinary. Water’s low expansibility reflects the fact that chemical bonds cannot be stretched by temperature. There is also a recent perturbation approach showing that it is more costly to accommodate a cavity of molecular size in water than in hexane as example.95 Considering the high fractional free volume for water (Table 13.1.2), it is concluded that the holes in water are distributed in smaller packets.96 Compared to a H-bonding network, a hard-sphere liquid finds more ways to configure its free volume in order to make a cavity. In the cavity-based model, large perturbations in water structure are not required to explain hydrophobic behavior. This conclusion arose out of the surprising success of the scaled particle theory (SPT),39 which is a hard-sphere fluid theory, to account for the free energy of hydrophobic transfers. Since the theory only uses the molecular size, density, and pressure of water as inputs and does not explicitly include any special features of H-bonding of water, the structure of water is arguably not directly implicated in the hydration thermodynamics. (However, the effect of H-bonds of water is implicitly taken into account through the size and density of water.) The proponents of this hypothesis argue that the entropic and enthalpic contributions arising from the structuring of water molecules largely compensate each other. In fact, there is thermodynamic evidence of enthalpy-entropy compensation of solvent reorganization.97-100 Furthermore, recent simulations101,102 and neutron scattering data103-105 suggest that solvent structuring might be of much lower extent than previously believed. Also a recent MD study report106 stated that the structure of water is preserved, rather than enhanced, around hydrophobic groups. Finally, the contribution of solute-water correlations to the hydrophobic effect may be displayed, for example, in the framework of the equation

760

Roland Schmid

∆Gsol = ∆G cav + ∆G att

[13.1.18]

where: ∆Gsol ∆Gcav ∆Gatt

free energy of dissolution free energy of cavity formation free energy of attractive interactions

This equation has been used by de Souza and Ben-Amotz107 to calculate values of ∆Gatt from the difference between experimental solubilities of rare gases, corresponding to ∆Gsol, and ∆Gcav assessed from eqn. [13.1.10], i.e., using a hard-sphere fluid (HF) model. The values of ∆Gatt so obtained have been found to correlate with the solute polarizabilities suggesting a dispersive mechanism for attractive solvation. It is interesting to note that, in water, the solubility of the noble gases increases with increasing size, in contrast to the aliphatic hydrocarbons whose solubility decreases with size. This differential behavior is straightforwardly explained in terms of the high polarizability of the heavy noble gases having a large number of weakly bound electrons, which strengthens the vdW interactions with water. It can be shown that for noble gases, on increasing their size, the vdW interactions increase more rapidly than the work of cavity creation, enhancing solubility. On the contrary, for the hydrocarbons, on increasing the size, the vdW interactions increase less rapidly than the work of cavity creation, lowering the solubility.108 We have seen that there is evidence of either model, the clathrate cage model and the cavity-based model. Hence the importance of water structure enhancement in the hydrophobic effect is equivocal. The reason for this may be twofold. First, theoretical models have many adjustable parameters, so their physical bases are not always clear. Second, the free energy alone masks the underlying physics in the absence of a temperature dependence study, because of, amongst other things, the entropy-enthalpy compensation noted above. In place of the free energy, other thermodynamic derivatives are more revealing. Of these, the study of heat capacity changes arguably provides a better insight into the role of changes in water structure upon hydration than analysis of entropy or enthalpy changes alone. Note that heat capacity is the most complex of the four principal thermodynamic parameters describing solvation (∆G, ∆H, ∆S, ∆Cp), with the following connections, ∆Cp =

∂∆H ∂∆S ∂ 2 ∆G =T = −T 2 ∂T ∂T ∂T 2

[13.1.19]

It should be stressed that the negative entropy of hydration is virtually not the main characteristic feature of hydrophobicity, since the hydration of any solute, polar, nonpolar, or ionic, is accompanied by a decrease in entropy.109 The qualitative similarity in hydration entropy behavior of polar and nonpolar groups contrasts sharply with the opposite sign of the heat capacity change in polar and nonpolar group hydration. Nonpolar solutes have a large positive heat capacity of hydration, while polar groups have a smaller, negative one. Thus, the large heat capacity increase might be what truly distinguishes the hydrophobic effect from other solvation effects.110 Recently, this behavioral difference of nonpolar and polar solutes could be reproduced by heat capacity calculations using a combination of Monte Carlo simulations and the random network model (RNM) of water.110-112 It was found that the hydrogen bonds between the water molecules in the first hydration shell of a nonpolar solute are shorter and less bent (i.e., are more ice-like) compared to those in pure water. The opposite effect occurs around


761

polar solutes (the waters become less ice-like). The increase in H-bond length and angle has been found to decrease the water heat capacity contribution, while decreases in length and angle have been found to cause the opposite effect. Note further that a large heat capacity implies that the enthalpy and entropy are strong functions of temperature, and the free energy vs. temperature is a curved function, increasing at low temperatures and decreasing at higher temperatures. Hence there will be a temperature at which the solubility of nonpolar in water is a minimum. The low solubility of nonpolar species in water at higher temperatures is caused by unfavorable enthalpic interactions, not unfavorable entropy changes. Some light on these features has been shed by using a “simple” statistical mechanical MB model of water in which the water molecules are represented as Lennard-Jones disks with hydrogen bonding arms.113 (the MB model is called this because of the resemblance of each model water to the Mercedes-Benz logo.) As an important result, the insertion of a nonpolar solute into cold water causes ordering and strengthening of the H bonds in the first shell, but the reverse applies in hot water. This provides a physical interpretation for the crossover temperatures TH and TS, where the enthalpy and entropy of transfer equal zero. TH is the temperature at which H-bond reorganizations are balanced by solute-solvent interactions. On the other hand, TS is the temperature at which the relative H-bonding strengths and numbers of shell and bulk molecules reverse roles. Although the large positive free energy of mixing of hydrocarbons with water is dominated by entropy at 25°C, it is dominated by enthalpy at higher temperatures (112°C from Baldwin’s extrapolation for hydrocarbons, or 150°C from the measurements of Crovetto for argon)113 where the disaffinity of oil for water is maximal. Ironically so, where hydrophobicity is strongest, entropy plays no role. For this reason, models and simulations of solutes that focus on cold water, around or below 25°C, miss much of the thermodynamics of the oil/water solvation process. Also, a clathrate-like solvation shell emerged from a recent computer simulation study of the temperature dependence of the structural and dynamical properties of dilute O2 aqueous solutions.114 In the first hydration shell around O2, water-water interactions are stronger and water diffusional and rotational dynamics slower than in the bulk. This calls to one’s mind an older paper by Hildebrand115 showing that at 25°C, methane’s diffusion coefficient in water is 40% less than it is in carbon tetrachloride (D(H2O) = 1.42x10-5 cm2/s vs D(CCl4) = 2.89x10-5 cm2/s). Presumably the loose clathrate water cages serve to inhibit free diffusion of the nonpolar solute. From these data it seems that both the nonpolar solute and the aqueous solvent experience a decrease in entropy upon dissolving in water. It should also be mentioned in this context that pressure increases the solubility. The effect of pressure on the entropy was examined and it was found that increase in the pressure causes a reduction of orientational correlations, in agreement with the idea of pressure as a “structure breaker” in water.116 Actually, frozen clathrate hydrates trapped beneath oceans and arctic permafrost may contain more than 50% of the world’s organic carbon reserves.117,118 Likewise, the solubility of aromatics is increased at high pressure and temperature, with π bond interactions involved.119 Only at first glance, the two approaches, the clathrate cage model and the cavity-based model, looked very different, the former based on the hydrogen bonding of water, and the later on the hard core of water. But taken all results together it would appear that both are just different perspectives on the same physics with different diagnostics reporting consequences of the same shifted balance between H bonds and vdW interactions. Actually, in a

762

Roland Schmid

very recent paper, a unified physical picture of hydrophobicity based on both the hydrogen bonding of water and the hard-core effect has been put forward.120 Hydrophobicity features an interplay of several factors. The structure of liquids A topic of abiding interest is the issue of characterizing the order in liquids which may be defined as the entropy deficit due to preferential orientations of molecular multipoles relative to random orientations (orientational order) and nonuniformly directed intermolecular forces (positional order). Phenomenologically, two criteria are often claimed to be relevant for deciding whether or not a liquid is to be viewed as ordered: the Trouton entropy of vaporization or Trouton quotient and the Kirkwood correlation factor gK.121 Strictly speaking, however, both are of limited relevance to the issue. The Trouton quotient is related to structure of the liquids only at their respective boiling points, of course, which may markedly differ from their structures at room temperature. This should be realized especially for the high-boiling liquids. As these include the highly dipolar liquids such as HMPA, DMSO, and PC, the effect of dipole orientation to produce order in the neat liquids remains obscure. All that can be gleaned from the approximate constancy of the Trouton quotient for all sorts of aprotic solvents is that at the boiling point the entropy of attractions becomes unimportant relative to the entropy of unpacking liquid molecules, that is repulsions.122 In terms of the general concept of separating the interaction potential into additive contributions of repulsion and attraction, the vaporization entropy can be expressed by ∆v H / Tb = ∆v S = ∆S o − S att

[13.1.20]

where: ∆ vH ∆So Satt

vaporization enthalpy at the boiling point Tb entropy of depacking hard spheres = entropy of repulsion entropy of attraction

∆So can be separated into the entropy fo(η) of depacking hard spheres to an ideal gas at the liquid volume V, and the entropy of volume expansion to Vg = RT/P, ∆S o / R = f o (η) + ln(Vg / V )

[13.1.21]

Further, fo(η) can be derived from the famous Charnahan-Starling equation as43

(

)

f o (η) = 4η − 3η 2 / (1 − η)

2

[13.1.22]

It can in fact be shown that, for the nonpolar liquids, ∆vS/R is approximately equal to fo(η).122 Along these lines, Trouton’s rule is traced to two facts: (i) The entropy of depacking is essentially constant, a typical value being ∆So/R ≈ 9.65, due to the small range of packing densities encompassed. In addition, the entropies of HS depacking and of volume change vary in a roughly compensatory manner. (ii) The entropy of attraction is insignificant for all the aprotics. Only for the protics the contributions of Satt may not be neglected. Actually the differences between ∆vS and ∆So reflect largely the entropy of hydrogen bonding. However, the application of the eqns [13.1.20] to [13.1.22] to room temperature data reveals, in con-


763

trast to boiling point conditions, not unimportant contributions of Satt even for some aprotic liquids (see below). On the other hand, the gK factor is, loosely speaking, a measure of the deviation of the relative dielectric constant of the solvent with the same dipole moment and polarizability would have if its dipoles were not correlated by its structure. However, the gK factor is only the average cosine of the angles between the dipole moments of neighboring molecules. There may thus be orientational order in the vicinity of a molecule despite a gK of unity if there are equal head-to-tail and antiparallel alignments. Furthermore, the gK factor is not related to positional order. The better starting point for assessing order would be experimental room temperature entropies of vaporization upon applying the same method as described above for the boiling point conditions. (Note however, that the packing density, and hence the molecular HS diameter, varies with temperature. Therefore, in the paper122 the packing densities have been calculated for near boiling point conditions.) Thus we choose the simplest fluid as the reference system. This is a liquid composed of spherical, nonpolar molecules, approximated to a HS gas moving in a uniform background or mean field potential provided by the attractive forces.43 Since the mean field potential affects neither structure nor entropy, the excess entropy Sex S ex / R = ln(Vg / V ) + f o (η) − ∆v H / RT

[13.1.23]

may be viewed as an index of orientational and positional order in liquids. It represents the entropy of attractions plus the contributions arising from molecular nonsphericity. This latter effect can be estimated by comparing the entropy deficits for spherical and hard convex body repulsions in the reference system. Computations available for three n-alkanes suggest that only ≈ 20% of Sex are due to nonsphericity effects. Table 13.1.6. Some liquid properties concerning structure. Data are from ref. 55 and 128 (gK) ∆vH/RT

∆vSo/R

-Sex/R

c-C6

18.21

17.89

0.32

THF

12.83

12.48

0.35

CCl4

13.08

12.72

0.36

n-C5

10.65

10.21

0.44

CHCl3

12.62

12.16

0.46

CH2Cl2

11.62

11.14

0.48

Ph-H

13.65

13.08

0.57

Ph-Me

15.32

14.62

0.70

n-C6

12.70

11.89

0.81

Et2O

10.96

10.13

0.83

c-hexanone

18.20

17.30

0.90

Solvent

gK

764

Roland Schmid

∆vH/RT

∆vSo/R

-Sex/R

Py

16.20

15.22

0.98

MeCN

13.40

12.37

1.03

1.18

Me2CO

12.50

11.40

1.10

1.49

MeOAc

13.03

11.79

1.24

EtCN

14.53

13.25

1.28

1.15

PhNO2

22.19

20.89

1.30

1.56

MeNO2

15.58

14.10

1.48

1.38

NMP

21.77

20.25

1.52

1.52

EtOAc

14.36

12.76

1.60

DMSO

21.33

19.30

2.03

1.67

DMF

19.19

17.04

2.15

1.60

DMA

20.26

18.04

2.22

1.89

HMPA

24.65

22.16

2.49

1.44

Solvent

gK

PhCN

21.97

19.41

2.56

NMF

22.69

19.75

2.94

n-C11

22.76

19.66

3.10

FA

24.43

21.26

3.17

2.04

H 2O

17.71

14.49

3.22

2.79

MeOH

15.10

11.72

3.38

2.99

n-C13

26.72

22.51

4.21

EtOH

17.07

12.85

4.22

3.08

PC

26.33

22.06

4.27

1.86

n-PrOH

19.14

14.29

4.85

3.23

n-BuOH

21.12

15.90

5.22

3.26

4.52

The calculations for some common liquids are given in Table 13.1.6 ordered according to decreasing Sex. An important result is the appreciable order produced by the hydrocarbon chain relative to polar groups and hydrogen-bonding effects. For instance, Sex would project for water the same degree of order as for undecane. In like terms, ethanol is comparable to tridecane. However, the same magnitude of the excess entropy does by no means imply that ordering is similar. Much of the orientational ordering in liquids composed of elongated molecules is a consequence of efficient packing such as the intertwining of chains. In contrast, the structure of water is largely determined by strong electrostatic interactions leading to sharply-defined directional correlations characteristic of H-bonding. Although contrary to chemical tradition, there are other indications that the longer-chain hydrocarbon liquids are to be classified as highly structured as judged from


765

thermodynamic123,124 and depolarized Rayleigh scattering data,125,126 and vibrational spectra.127 For nonprotic fluids, as Table 13.1.6 further shows, the vaporization entropy is strongly dominated by the entropy of HS depacking. This is an at least qualitative representation of the longstanding claim that repulsions play the major role in the structure of dense fluids.37 This circumstance is ultimately responsible for the striking success of the description of neutral reactions in the framework of a purely HS liquid, as discussed in Section 13.1.6. Also included in the Table are values of gK as determined in the framework of a generic mean spherical approximation.128 Since these values differ from those from other sources,129,121 because of differences in theory, we refrain from including the latter. It is seen that the gK parameter is unsuited to scale order, since positional order is not accounted for. On the other hand, values of gK exceed unity for the highly dipolar liquids and thus both Sex and gK attest to some degree of order present in them. Solvent reorganization energy in ET Electron transfer (ET) reactions in condensed matter continue to be of considerable interest to a wide range of scientists. The reasons are twofold. Firstly, ET plays a fundamental role in a broad class of biological and chemical processes. Secondly, ET is rather simple and very suitable to be used as a model for studying solvent effects and to relate the kinetics of ET reactions to thermodynamics. Two circumstances make ET reactions particularly appealing to theoreticians: • Outer-sphere reactions and ET within rigid complexes of well-defined geometry proceed without changes in the chemical structure, since bonds are neither formed nor broken. • The long-ranged character of interactions of the transferred electron with the solvent’s permanent dipoles. As a consequence of the second condition, a qualitative (and even quantitative) description can be achieved upon disregarding (or reducing through averaging) the local liquid structure changes arising on the length of molecular diameter dimensions relative to the charge-dipole interaction length. Because of this, it becomes feasible to use for outer-sphere reactions in strongly polar solvents the formalism first developed in the theory of polarons in dielectric crystals.130 In the treatment, the polar liquid is considered as a dielectric continuum characterized by the high-frequency ε ∞ and static εs dielectric constants, in which the reactants occupy spherical cavities of radii Ra and Rd, respectively. Electron transitions in this model are supposed to be activated by inertial polarization of the medium attributed to the reorientation of permanent dipoles. Along these lines Marcus131 obtained his well-known expression for the free energy ∆F of ET activation ∆F =

(∆Fo

+Er)

4E r

2

[13.1.24]

where ∆Fo is the equilibrium free energy gap between products and reactants and Er is the reorganization energy equal to the work applied to reorganize inertial degrees of freedom changing in going from the initial to the final charge distribution and can be dissected into inner-sphere and solvent contributions:

766

Roland Schmid

Er = Ei + Es

[13.1.25]

where: Ei Es

inner-sphere reorganization energy solvent reorganization energy

For outer-sphere ET the solvent component Es of the reorganization energy Es = e2cog

[13.1.26]

is the product of the medium-dependent Pekar factor co = 1/ε ∞ - 1/εs and a reactant-dependent (but solvent independent) geometrical factor g = 1/2Ra + 1/2Rd - 1/R

[13.1.27]

where: R e

the donor-acceptor separation the electron charge

Further advancements included calculations of the rate constant preexponent for nonadiabatic ET,132 an account of inner-sphere133 and quantum intramolecular134-136 vibrations of reactants and quantum solvent modes.137,138 The main results were the formulation of the dependencies of the activation energy on the solvent dielectric properties and reactant sizes, as well as the bell-shaped relationship between ∆F and ∆Fo. The predicted activation energy dependence on both the solvent dielectric properties139,140 and the donor-acceptor distance141 has, at least qualitatively, been supported by experiment. A bell-shaped plot of ∆F vs ∆Fo was obtained for ET in exciplexes,142 ion pairs,143 intramolecular144 and outer-sphere145 charge shift reactions. However, the symmetric dependence predicted by eq. [13.1.24] has not yet been detected experimentally. Instead, always asymmetric plots of ∆F against ∆Fo are obtained or else, in the inverted region (∆Fo < -Er), ∆F was found to be nearly invariant with ∆Fo.146 A couple of explanations for the asymmetric behavior are circulating in the literature (see, e.g., the review by Suppan147). The first one148,149 considered vibrational excitations of high-frequency quantum vibrational modes of the donor and acceptor centers. Another suggestion150 was that the frequencies of the solvent orientational mode are significantly different around the charged and the neutral reactants. This difference was supposed to be brought about by dielectric saturation of the polar solvent. This model is rightly questioned151 since dielectric saturation cannot affect curvatures of the energy surface at the equilibrium point. Instead, dielectric saturation displays itself in a nonlinear deviation of the free energy surface from the parabolic form far from equilibrium. Hence, other sources of this behavior should be sought. Nevertheless, both concepts tend to go beyond the structureless description advocating a molecular nature of either the donor-acceptor complex or the solvent. Nowadays, theories of ET are intimately related to the theories of optical transitions. While formerly both issues have developed largely independently, there is now growing desire to get a rigorous description in terms of intermolecular forces shifting the research of ET reactions toward model systems amendable to spectroscopic methods. It is the combination of steady state and transient optical spectroscopy that becomes a powerful method of studying elementary mechanisms of ET and testing theoretical concepts. The classical treatments of ET and optical transition have been facing a serious problem when extended to weakly polar and eventually nonpolar solvents. Values of Eop (equal to Er in eq [13.1.25]) as


767

extracted from band-shape analyses of absorption spectra were found to fall in the range 0.2 - 0.4 eV. Upon partitioning these values into internal vibrations and solvent degrees of freedom, although this matter is still ambiguous, the contribution of the solvent could well be on the order of 0.2 - 0.3 eV.152-154 Unfortunately, all continuum theories predict zero solvent reorganization energies for ET in nonpolar liquids. It is evident that some new mechanisms of ET, alternatively to permanent dipoles’ reorientation, are to be sought. It should be emphasized that the problem cannot be resolved by treatments of fixed positions of the liquid molecules, as their electronic polarization follows adiabatically the transferred electron and thus cannot induce electronic transitions. On the other hand, the displacement of molecules with induced dipoles are capable of activating ET. In real liquids, as we have stated above, the appreciable free volume enables the solvent molecules to change their coordinates. As a result, variations in charge distribution in the reactants concomitantly alter the packing of liquid molecules. This point is corroborated by computer simulations.155 Charging a solute in a Stockmayer fluid alters the inner coordination number from 11.8 for the neutral entity to 9.5 for the positively charged state, with the process accompanied by a compression of the solvation shell. It is therefore apparent that solvent reorganization involves reorganization of liquid density, in addition to the reorientational contribution. This concept has been introduced by Matyushov,156 who dissected the overall solvent reorganization energy Es into a dipole reorganization component Ep and a density reorganization component Ed, Es = Ep + Ed

[13.1.28]

It should be mentioned that the two contributions can be completely separated because they have different symmetries, i.e., there are no density/orientation cross terms in the perturbation expansion involved in the calculations. The density component comprises three mechanisms of ET activation: (i) translations of permanent dipoles, (ii) translations of dipoles induced by the electric field of the donor-acceptor complex (or the chromophore), and (iii) dispersion solute-solvent forces. On the other hand, it appears that in the orientational part only the permanent dipoles (without inductions) are involved. With this novel molecular treatment of ET in liquids the corundum of the temperature dependence of the solvent reorganization energy is straightforwardly resolved. Dielectric continuum theories predict an increase of Es with temperature paralleling the decrease in the dielectric constants. In contrast, experimental results becoming available quite recently show that Es decreases with temperature. Also curved Arrhenius plots eventually featuring a maximum are being reported, in weakly polar157 and nonpolar158 solvents. The bell-shaped temperature dependence in endergonic and moderately exergonic regions found for ET quenching reactions in acetonitrile159 was attributed to a complex reaction mechanism. Analogously, the maximum in the Arrhenius coordinates, peculiar to the fluorescence of exciplexes formed in the intramolecular160 and bimolecular161 pathways, is commonly attributed to a temperature dependent competition of exciplex formation and deactivation rates. A more reasonable explanation can be given in terms of the new theory as follows. A maximum in the Arrhenius coordinates follows from the fact that the two terms in eq. [13.1.28] depend differently on temperature. Density fluctuation around the reacting pair is determined mainly by the entropy of repacking hard spheres representing the repulsive part of the intermolecular interaction. Mathematically, the entropy of activation arises

768

Roland Schmid

from the explicit inverse temperature dependence Ed ∝1/T. Since the liquid is less packed at higher temperature, less energy is needed for reorganization. Repacking of the solvent should lead to larger entropy changes than those of dipoles’ reorientation that is enthalpic in nature due to the long-range character of dipole-dipole forces. The orientational component increases with temperature essentially as predicted by continuum theories. In these ways the two solvent modes play complementary roles in the solvent’s total response. This feature would lead to curved Arrhenius plots of ET rates with slight curvatures in the normal region of ET (-∆Fo < Er), but even a maximum in the inverted region (-∆Fo > Er).162,163 The maximum may however be suppressed by intramolecular reorganization and should therefore be discovered particularly for rigid donor-acceptor pairs. Photoinduced ET in binuclear complexes with localized electronic states provides at the moment the best test of theory predictions for the solvent dependent ET barrier. This type of reaction is also called metal-metal charge-transfer (MMCT) or intervalence transfer (IT). The application of the theory to IT energies for valence localized biruthenium complexes and the acetylene-bridged biferricenium monocation164 revealed its superiority to continuum theories. The plots of Es vs. Eop are less scattered, and the slopes of the best-fit lines are closer to unity. As a major merit, the anomalous behavior of some solvents in the continuum description - in particular HMPA and occasionally water - becomes resolved in terms of the extreme sizes, as they appear at the opposite ends of the solvent diameter scale. Recently, it became feasible for the first time, to measure experimentally for a single chemical system, viz. a rigid, triply linked mixed-valence binuclear iron polypyridyl complex, [Fe(440)3Fe]5+, the temperature dependencies of both the rate of thermal ET and the optical IT energy (in acetonitrile-d3).165 The net Er associated with the intramolecular electron exchange in this complex is governed exclusively by low frequency solvent modes, providing an unprecedented opportunity to compare the parameters of the theories of thermal and optical ET in the absence of the usual complications and ambiguities. Acceptable agreement was obtained only if solvent density fluctuations around the reacting system were taken into account. In these ways the idea of density fluctuations is achieving experimental support. The two latest reports on negative temperature coefficients of the solvent reorganization energy (decrease in Es with temperature) should also be mentioned.166,167 Thus, two physically important properties of molecular liquids are absent in the continuum picture: the finite size of the solvent molecules and thermal translational modes resulting in density fluctuations. Although the limitations of the continuum model are long known, the necessity for a molecular description of the solvent, curiously, was first recognized in connection with solvent dynamic effects in ET. Solvent dynamics, however, affects the preexponent of the ET rate constant and, therefore, influences the reaction rate much less than does the activation energy. From this viewpoint it is suspicious that the ET activation energy has so long been treated in the framework of continuum theories. The reason of this affection is the otherwise relative success of the latter, traceable to two main features. First, the solvents usually used are similar in molecular size. Second, there is a compensation because altering the size affects the orientational and translational parts of the solvent barrier in opposite directions. The solution ionic radius The solution ionic radius is arguably one of the most important microscopic parameters. Although detailed atomic models are needed for a full understanding of solvation, simpler phenomenological models are useful to interpret the results for more complex systems. The


769

most famous model in this respect is that of Born, originally proposed in 1920,168 representing the simplest continuum theory of ionic solvation. For a spherical ion, the Born excess free energy ∆GB of solvation was derived by considering the free energy change resulting from the transfer of an ion from vacuum to solvent. The equation has a very simple dependence on the ionic charge z, the radius rB, and the solvent dielectric constant ε (for the prime see eq. [13.1.16]): ∆G *B =

−e 2 z 2  1   1−  2rB  ε 

[13.1.29]

While Born assumes that the dielectric response of the solvent is linear, nonlinear effects such as dielectric saturation and electrostriction should occur due to the high electric field near the ion.169 Dielectric saturation is the effect that the dipoles are completely aligned in the direction of the field so that any further increase in the field cannot change the degree of alignment. Electrostriction, on the other hand, is defined as the volume change or compression of the solvent caused by an electric field, which tends to concentrate dipoles in the first solvation shell of an ion. Dielectric saturation is calculated to occur at field intensities exceeding 104 V/cm while the actual fields around monovalent ions are on the order of 108 V/cm.170 In the following we concentrate on ionic hydration that is generally the focus of attention. Unaware of nonlinear effects, Latimer et al.171 showed that the experimental hydration free energies of alkali cations and halide anions were consistent with the simple Born equation when using the Pauling crystal radii rP increased by an empirical constant ∆ equal to 0.85 Å for the cations and 0.1 Å to the anions. In fact three years earlier a similar relationship was described by Voet.172 The distance rP + ∆ was interpreted as the radius of the cavity formed by the water dipoles around the ion. For cations, it is the ion-oxygen distance while for anions it is the ion-hydrogen distance of the neighboring water molecules. From those days onwards, the microscopic interpretation of the parameters of the Born equation has continued to be a corundum because of the ambiguity of using either an effective radius (that is a modification of the crystal radii) or an effective dielectric constant. Indeed, the number of modifications of the Born equation is hardly countable. Rashin and Honig,173 as example, used the covalent radii for cations and the crystal radii for anions as the cavity radii, on the basis of electron density distributions in ionic crystals. On the other hand, Stokes174 put forward that the ion’s radius in the gas-phase might be appreciably larger than that in solution (or in a crystal lattice of the salt of the ion). Therefore, the loss in self-energy of the ion in the gas-phase should be the dominant contributor. He could show indeed that the Born equation works well if the vdW radius of the ion is used, as calculated by a quantum mechanical scaling principle applied to an isoelectronic series centering around the crystal radii of the noble gases. More recent accounts of the subject are available.175,176 Irrespective of these ambiguities, the desired scheme of relating the Born radius with some other radius is facing an awkward situation: Any ionic radius depends on arbitrary divisions of the lattice spacings into anion and cation components, on the one hand, and on the other, the properties of individual ions in condensed matter are derived by means of some extra-thermodynamic principle. In other words, both properties, values of r and ∆G*, to be compared with one another, involve uncertain apportionments of observed quantities. Con-

770

Roland Schmid

sequently, there are so many different sets of ionic radii and hydration free energies available that it is very difficult to decide which to prefer. In a most recent paper,82 a new table of absolute single-ion thermodynamic quantities of hydration at 298 K has been presented, based on conventional enthalpies and entropies upon implication of the thermodynamics of water dissociation. From the values of ∆hydG* the Born radii were calculated from rB (Å) = -695 z2/∆hydG*(kJ)

[13.1.30]

as given in Table 13.1.7. This is at first a formal definition whose significance may be tested in the framework of the position of the first maximum of the radial distribution function (RDF) measured by solution X-ray and neutron diffraction.177 However, the procedure is not unambiguous as is already reflected by the names given to this quantity, viz. (for the case of a cation) ion-water178 or ion-oxygen Table 13.1.7. Some radii (Å). Data are from distance. The ambiguity of the underlying ref. 82 interpretation resides in the circumstance that the same value of 1.40 Å is assigned in a b raq rmetal Atom rB the literature to the radius of the oxide anion, the water molecule and the vdW radius Li 1.46 1.50 1.52 of the oxygen atom. Na 1.87 1.87 1.86 It seems that many workers would K 2.33 2.32 2.27 tend to equate the distance (d) corresponding to the first RDF peak with the average Rb 2.52 2.48 distance between the center of the ion and Cs 2.75 2.58 2.65 the centers of the nearest water molecules, Be 1.18 1.06 1.12 d=rion + rwater. Actually, Marcus179,180 presented a nice relationship between d, averMg 1.53 1.52 1.60 aged over diffraction and simulation data, Ca 1.86 1.86 1.97 and the Pauling crystal radius in the form Sr 2.03 2.02 2.15 d=1.38 + 1.102 rp. Notwithstanding this success, it is preferable to implicate not the waBa 2.24 2.27 2.17 ter radius but instead the oxygen radius. F 1.39 1.29 This follows from the close correspondence between d and the metal-oxygen bond Cl 1.86 1.85 lengths in crystalline metal hydrates.82 Br 2.00 2.00 The gross coincidence of the solid and I 2.23 2.30 solution state distances is strong evidence that the value of d measures the distance bea From eqn. [13.1.30], beqns. [13.1.31] and [13.1.32]. tween the nuclei of the cation ad the oxygen rather than the center of the electron cloud of the whole ligand molecule. Actually, first RDF peaks for ion solvation in water and in nonaqueous oxygen donor solvents are very similar despite the different ligand sizes. Examples include methanol, formamide and dimethyl sulfoxide.180 Nevertheless, the division of d into ion and ligand components is still not unequivocal. Since the traditional ionic radius is often considered as a literal measure of size, it is usual to


771

interpret crystallographic metal-oxygen distances in terms of the sum of the vdW radius of oxygen and the ionic radius of the metal. It should b emphasized, however, that the division of bond lengths into “cation” and “anion” components is entirely arbitrary. If the ionic radius is retained, the task remains to seek a connection to the Born radius, an issue that has a long-standing history beginning with the work of Voet.172 Of course, any addition to the ionic radius necessary to obtain good results from the Born equation needs a physical explanation. This is typically done in terms of the water radius, in addition to other correction terms such as a dipolar correlation length in the MSA (mean spherical approximation).181-183 In this case, however, proceeding from the first RDF peak, the size of the water moiety is implicated twice. It has been shown82 that the puzzle is unraveled if the covalent (atomic) radius of oxygen is subtracted from the experimental first peak position of the cation-oxygen radial distribution curve (strictly, the upper limits instead of the averages). The values of raq so obtained are very close to the Born radius, d(cation-O) - rcov(O) = raq ≈ rB

[13.1.31]

Similarly, for the case of the anions, the water radius, taken as 1.40 Å, is implicated, d(anion-O) - r(water) = raq ≈ rB

[13.1.32]

Furthermore, also the metallic radii (Table 13.1.7) are similar to values of raq. This correspondence suggests that the positive ion core dimension in a metal tends to coincide with that of the corresponding rare gas cation. The (minor) differences between raq and rmetal for the alkaline earth metals may be attributed, among other things, to the different coordination numbers (CN) in the metallic state and the solution state. The involvement of the CN is apparent in the similarity of the metallic radii of strontium and barium which is obviously a result of cancellation of the increase in the intrinsic size in going from Sr to Ba and the decrease in CN from 12 to 8. Along these lines a variety of radii are brought under one umbrella, noting however a wide discrepancy to the traditional ionic radii. Cation radii larger than the traditional ionic radii would imply smaller anion radii so as to meet the (approximate) additivity rule. In fact, the large anion radii of the traditional sets give rise to at least two severe inconsistencies: (i) The dramatic differences on the order of 1 Å between the covalent radii and the anion radii are hardly conceivable in view of the otherwise complete parallelism displayed between ionic and covalent bonds.184 (ii) It is implausible that non-bonded radii185 should be smaller than ionic radii. For example, the ionic radius of oxygen of 1.40 Å implies that oxygen ions should not approach each other closer than 2.80 Å. However, non-bonding oxygen-oxygen distances as short as 2.15 Å have been observed in a variety of crystalline environments. “(Traditional) ionic radii most likely do not correspond to any physical reality,” Baur notes.186 It should be remarked that the scheme of reducing the size of the anion at the expense of that of the cation has been initiated by Gourary and Adrian, based on the electron density contours in crystals.187 The close correspondence seen between rB and raq supports the idea that the Born radius (in aqueous solution) is predominantly a distance parameter without containing dielectric, i.e., solvent structure, contributions. This result could well be the outcome of a cancellation of dielectric saturation and electrostriction effects as suggested recently from

772

Roland Schmid

simulations.188-191 It should be emphasized, however, that the present discussion might be confined to water as the solvent. Recent theoretical treatments advise the cavity radius not to be considered as an intrinsic property of the solute, but instead to vary with solvent polarity, with orientational saturation prevailing at low polarity and electrostriction at high polarity.191,192 It would appear that the whole area of nonaqueous ion solvation deserves more methodical attention. It should in addition be emphasized that the cavity radius is sensitive to temperature. Combining eq. [13.1.29] with  ∂∆G *  ∆H o = ∆G * +T    ∂T 

[13.1.33]

one obtains177  T  ∂ε  T  ∂r   ∆H o = ∆G * 1 +     −   (ε − 1)ε  ∂T  P r  ∂T  P 

[13.1.34]

The derivation of (∂r/∂T)P from reliable values of ∆H° and ∆G* is interesting, in that nominally the dielectric effect (-0.018 for water) is smaller than the size effect (-0.069 for chloride), a result that has not given previously the attention due to it. Consequently, as Roux et al.177 stated, unless a precise procedure for evaluating the dependence of the radius on the temperature is available, the Born model should be restricted to the free energy of solvation. Notwithstanding this, beginning with Voet,172 the Born model has usually been tested by considering the enthalpies of hydration.193 The reason for the relative success lies in the fact that ion hydration is strongly enthalpy controlled, i.e., ∆H° ~ ∆G*. The discussion of radii given here should have implications to all calculations involving aqueous ionic radii, for instance the solvent reorganization energy in connection with eq. [13.1.26]. Thus, treatments using the crystal radii as an input parameter194-197 may be revisited. 13.1.8 THE FUTURE OF THE PHENOMENOLOGICAL APPROACH Originally, the empirical solvent parameters have been introduced to provide guidelines for the comparison of different solvent qualities and for an orientation in the search for an understanding of the complex phenomena in solution chemistry. Indeed, the choice of the right solvent for a particular application is an everyday decision for the chemist: which solvent should be the best to dissolve certain products, and what solvent should lead to increased reaction yields and/or rates of a reaction? In the course of time, however, a rather sophisticated scheme has developed of quantitative treatments of solute-solvent interactions in the framework of LSERs.198 The individual parameters employed were imagined to correspond to a particular solute-solvent interaction mechanism. Unfortunately, as it turned out, the various empirical polarity scales feature just different blends of fundamental intermolecular forces. As a consequence, we note at the door to the twenty-first century, alas with melancholy, that the era of combining empirical solvent parameters in multiparameter equations, in a scientific context, is beginning to fade away. As a matter of fact, solution chemistry research is increasingly being occupied by theoretical physics in terms of molecular dynamics (MD) and Monte Carlo (MC) simulations, the integral equation approach, etc.


773

In the author’s opinion, it seems that further usage of the empirical parameters should more return to the originally intended purpose, emphasizing more the qualitative aspects rather than to devote too much effort to multilinear regression analyses based on parameters quoted to two decimal places. Admittedly, such a scheme may nevertheless still be used to get some insight concerning the nature of some individual solvent effect as in the recent case of an unprecedented positive wavelength shift in the solvatochromism of an aminobenzodifuranone.199 The physical approach, though still in its infancy, has been helping us to see the success of the phenomenological approach in a new light. Accordingly, the reason for this well documented and appreciated success can be traced back to the following features • The molecular structure and the molecular size of many common solvents are relatively similar. The majority belongs to the so-called select solvents having a single dominant bond dipole, which, in addition, is typically hard, viz. an O- or N-donor. For example, if also soft donors (e.g., sulfur) had been employed to a larger extent, no general donor strength scale could have been devised. Likewise, we have seen that solvents other than the select ones complicate the issue. • As it runs like a thread through the present treatment, various cancellations and competitions (enthalpy/entropy, repulsion/attraction, etc.) appear to be conspiring to make molecular behavior in complex fluids seemingly simple. Notwithstanding this, the phenomenological approach will remain a venerable cornerstone in the development of unraveling solvent effects. Only time will tell whether a new generation of solvent indices will arise from the physical approach. REFERENCES 1

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

The reaction studied was the quaternarization of triethylamine by ethyl iodide at 100 °C [N. Menschutkin, Z. Phys. Chem., 6, 41 (1890)]. Menschutkin’s first discussion on solvent effects dealt with the reactions between acetic anhydride and alcohols [Z. Phys. Chem., 1, 611 (1887)]. The catalytic role of solvents was already recognized in 1862 by Berthelot and Péan de Saint Gilles in their Recherches sur les Affinités [see, e. g., H. G. Grimm, H. Ruf, and Wolff, Z. Phys. Chem., B13, 301 (1931)]. E. D. Hughes and C. K. Ingold, J. Chem. Soc., 244 (1935). H. Suhr, Ber. Bunsenges. Phys. Chem., 67, 893 (1963). R. Schmid, J. Solution Chem., 12, 135 (1983). V. Gutmann, Donor-Acceptor Approach to Molecular Interactions, Plenum Press, New York, 1978. N. V. Sidgwick, The Electronic Theory of Valency, Clarenton Press, Oxford, 1927. R. Schmid, Rev. Inorg. Chem., 11, 255 (1991). R. W. Taft, N. J. Pienta, M. J. Kamlet, and E. M. Arnett, J. Org. Chem., 46, 661 (1981). W. Linert and R. F. Jameson, J. Chem. Soc. Perkin Trans. 2, 1993, 1415. V. Bekárek, J. Phys. Chem., 85, 722 (1981). Y. Marcus, Chem. Soc. Rev., 409 (1993). C. Laurence, P. Nicolet, M. T. Dalati, J. M. Abboud, and R. Notario, J. Phys. Chem., 98, 5807 (1994). C. Reichardt, Chem. Rev., 94, 2319 (1994). M. R. J. Dack, Aust. J. Chem., 28, 1643 (1975). M. J. Kamlet, J. M. Abboud, M. H. Abraham, and R. W. Taft, J. Org. Chem., 48, 2877 (1983). R. W. Taft, M. H. Abraham, R. M. Doherty, and M. J. Kamlet, J. Am. Chem. Soc., 107, 3105 (1985). M. H. Abraham, P. L. Grellier, A. Nasehzadeh, and R. A. C. Walker, J. Chem. Soc. Perkin Trans. II, 1988, 1717. R. W. Soukup and R. Schmid, J. Chem. Educ., 62, 459 (1985). U. Mayer, Pure Appl. Chem., 51, 1697 (1979). M. J. Kamlet, J. L. M. Abboud, and R. W. Taft, Prog. Phys. Org. Chem., 13, 485 (1981). R. Schmid and V. N. Sapunov, Non-Formal Kinetics in Search for Chemical Reaction Pathways, Verlag Chemie, Weinheim, 1982. M. H. Abraham, R. M. Doherty, M. J. Kamlet, and R. W. Taft, Chem. Br., 22, 551 (1986).

774

23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75

Roland Schmid N. A. Lewis, Y. S. Obeng, and W. L. Purcell, Inorg. Chem., 28, 3796 (1989). J. T. Hupp and M. J. Weaver, Inorg. Chem., 23, 3639 (1984). S. Glikberg and Y. Marcus, J. Solution Chem., 12, 255 (1983). Y. Marcus, M. J. Kamlet, and R. W. Taft, J. Phys. Chem., 92, 3613 (1988). C. J. Timpson, C. A. Bignozzi, B. P. Sullivan, E. M. Kober, and T. J. Meyer, J. Phys. Chem., 100, 2915 (1996). U. Mayer and V. Gutmann, Structure and Bonding, 12, 113 (1972). Y. Marcus, J. Solution Chem., 13, 599 (1984). Y. Kondo, M. Ohnishi, and N. Tokura, Bull. Chem. Soc. Japan, 45, 3579 (1972). H. F. Herbrandson and F. R. Neufeld, J. Org. Chem., 31, 1140 (1966). M. J. Kamlet, P. W. Carr, R. W. Taft, and M. H. Abraham, J. Am. Chem. Soc., 103, 6062 (1981). B. Widom, Science, 157, 375 (1967). J. A. Barker and D. Henderson, J. Chem. Educ., 45, 2 (1968). Y. Song, R. M. Stratt, and E. A. Mason, J. Chem. Phys., 88, 1126 (1988). S. Phan, E. Kierlik, M. L. Rosinberg, H. Yu, and G. Stell, J. Chem. Phys., 99, 5326 (1993). C. G. Gray and K. E. Gubbins, Theory of Molecular Fluids, Vol.1: Fundamentals, Clarendon Press, Oxford, 1984. L. K. Nash in Harvard Case Histories in Experimental Science, J. B. Conant, Ed., Harvard University Press, Harvard, 1950, p. 17. R. A. Pierotti, J. Phys. Chem., 67, 1840 (1963); Chem. Rev., 76, 717 (1976). E. Wilhelm and R. Battino, J. Chem. Phys., 55, 4021 (1971); J. Chem. Thermodyn., 3, 761 (1971); J. Chem. Phys., 58, 3558 (1973). D. V. Matyushov and R. Schmid, J. Chem. Phys., 104, 8627 (1996). D. Ben-Amotz and D. R. Herschbach, J. Phys. Chem., 94, 1038 (1990). R. Schmid and D. V. Matyushov, J. Phys. Chem., 99, 2393 (1995). D. Ben-Amotz and K. G. Willis, J. Phys. Chem., 97, 7736 (1993). T. Boublik, J. Chem. Phys., 53, 471 (1970). G. A. Mansoori, N. F. Carnahan, K. E. Starling, and T. W. Leland, Jr., J. Chem. Phys., 54, 1523 (1971). D. V. Matyushov and B. M. Ladanyi, J. Chem. Phys., 107, 5851 (1997). R. Ravi, L. E. S. Souza, and D. Ben-Amotz, J. Phys. Chem., 97, 11835 (1993). D. Ben-Amotz, J. Phys. Chem., 97, 2314 (1993). D. V. Matyushov and R. Schmid, J. Chem. Phys., 103, 2034 (1995). J. D. Weeks, D. Chandler, and H. C. Andersen, J. Chem. Phys., 54, 5237 (1971). J. A. Barker and D. Henderson, Rev. Mod. Phys., 48, 587 (1976). F. Cuadros, A. Mulero, and P. Rubio, J. Chem. Educ., 71, 956 (1994). D. V. Matyushov and R. Schmid, J. Chem. Phys., 105, 4729 (1996). D. V. Matyushov, R. Schmid, and B. M. Ladanyi, J. Phys. Chem. B, 101, 1035 (1997). L. Reynolds, J. A. Gardecki, S. J. V. Frankland, M. L. Horng, and M. Maroncelli, J. Phys. Chem., 100, 10337 (1996). D. V. Matyushov and G. A. Voth, J. Chem. Phys., 111, 3630 (1999). S. R. Mente and M. Maroncelli, J. Phys. Chem. B, 103, 7704 (1999). J. Catalán, J. Org. Chem., 62, 8231 (1997). B. C. Perng, M. D. Newton, F. O. Raineri, and H. L. Friedman, J. Chem. Phys., 104, 7177 (1996). O. M. Carbacos, C. J. Weinheimer, and J. M. Lisy, J. Chem. Phys., 108, 5151 (1998). O. M. Carbacos, C. J. Weinheimer, and J. M. Lisy, J. Chem. Phys., 110, 8429 (1999). S. G. Bratsch, J. Chem. Educ., 65, 34 (1988). W. C. Röntgen, Ann. Phys. u. Chem., 45, 91 (1892). W. Kauzmann, L’Eau Syst. Biol., Colloq. Int. C.N.R.S., 246, 63 (1975). See, for example: H. Endo, J. Chem. Phys., 72, 4324 (1980). M. Vedamuthu, S. Singh, and G. W. Robinson, J. Phys. Chem., 98, 2222 (1994). I. Ohmine, J. Chem. Phys., 99, 6767 (1995). S. W. Benson and E. D. Siebert, J. Am. Chem. Soc., 114, 4269 (1992). F. O. Libnau, J. Toft, A. A. Christy, and O. M. Kvalheim, J. Am. Chem. Soc., 116, 8311 (1994). H. Kistenmacher, G. C. Lie, H. Pople, and E. Clementi, J. Chem. Phys., 61, 546 (1974). C. H. Cho, S. Singh, and G. W. Robinson, J. Chem. Phys., 107, 7979 (1997). L. Bosio, S. H. Chen, and J. Teixeira, Phys. Rev., A27, 1468 (1983). G. D’Arrigo, G. Maisano, F. Mallamace, P. Migliardo, and F. Wanderlingh, J. Chem. Phys., 75, 4264 (1981). A. K. Soper and M. G. Phillips, Chem. Phys., 107, 47 (1986).


76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131

775

S. W. Benson, J. Am. Chem. Soc., 100, 5640 (1978). K. A. T. Silverstein, A. D. J. Haymet, and K. A. Dill, J. Am. Chem. Soc., 120, 3166 (1998). E. A. Jagla, J. Chem. Phys., 111, 8980 (1999). E. D. Isaacs, A. Shukla, P. M. Platzman, D. R. Hamann, B. Barbiellini, and C. A. Tulk, Phys. Rev. Lett., 82, 600 (1999). L. Pauling, J. Am. Chem. Soc., 57, 2680 (1935). G. Nadig, L. C. Van Zant, S. L. Dixon, and K. M. Merz, Jr., J. Am. Chem. Soc., 120, 5593 (1998). R. Schmid, A. M. Miah, and V. N. Sapunov, Phys. Chem. Chem. Phys., 2, 97 (2000). A. Ben-Naim, Hydrophobic Interactions, Plenum, New York, 1980. D. F. Evans and B. W. Ninham, J. Phys. Chem., 90, 226 (1986). C. Tanford, The Hydrophobic Effect: Formation of Micelles and Biological Membranes, Wiley, New York, 1973. K. A. Dill, Biochemistry, 29, 7133 (1990). O. K. Forrisdahl, B. Kvamme, and A. D. J. Haymet, Mol. Phys., 89, 819 (1996). H. Wang and A. Ben-Naim, J. Med. Chem., 39, 1531 (1996). H. S. Ashbaugh, E. W. Kaler, and M. E. Paulaitis, J. Am. Chem. Soc., 121, 9243 (1999). H. S. Frank and M. W. Evans, J. Chem. Phys., 13, 507 (1945). K. Mizuno, Y. Miyashita, Y. Shindo, and H. Ogawa, J. Phys. Chem., 99, 3225 (1995). E. V. Goldammer and H. G. Hertz, J. Phys. Chem., 74, 3734 (1970). R. Pottel and V. Kaatze, Ber. Bunsenges. Phys. Chem., 73, 437 (1969). R. Silveston and B. Kronberg, J. Phys. Chem., 93, 6241 (1989). M. Prévost, I. T. Oliveira, J. P. Kocher, and S. J. Wodak, J. Phys. Chem., 100, 2738 (1996). A. Pohorille and L. R. Pratt, J. Am. Chem. Soc., 112, 5066 (1990). E. Grunwald and C. Steel, J. Am. Chem. Soc., 117, 5687 (1995). H. A. Yu, and M. Karplus, J. Chem. Phys., 89, 2366 (1988). H. Qian and J. J. Hopfield, J. Chem. Phys., 105, 9292 (1996). B. Lee, J. Chem. Phys., 83, 2421 (1985). I. I. Vaisman, F. K. Brown, and A. Tropsha, J. Phys. Chem., 98, 5559 (1994). M. Re, D. Laria, and R. Fernández-Prini, Chem. Phys. Lett., 250, 25 (1996). A. K. Soper and J. L. Finney, Phys. Rev. Lett., 71, 4346 (1993). J. Turner and A. K. Soper, J. Chem. Phys., 101, 6116 (1994). A. K. Soper and A. Luzar, J. Phys. Chem., 100, 1357 (1996). E. C. Meng and P. A. Kollman, J. Phys. Chem., 100, 11460 (1996). L. E. S. de Souza and D. Ben-Amotz, J. Chem. Phys., 101, 9858 (1994). G. Graziano, J. Chem. Soc., Faraday Trans., 94, 3345 (1998). A. Ben-Naim and Y. Marcus, J. Chem. Phys., 81, 2016 (1984). B. Madan and K. Sharp, J. Phys. Chem., 100, 7713 (1996). K. Sharp and B. Madan, J. Phys. Chem. B, 101, 4343 (1997). B. Madan and K. Sharp, J. Phys. Chem. B, 101, 11237 (1997). K. A. T. Silverstein, A. D. J. Haymet, and K. A. Dill, , J. Chem. Phys., 111, 8000 (1999). E. Fois, A. Gamba, and C. Redaelli, J. Chem. Phys., 110, 1025 (1999). J. H. Hildebrand, Proc. Natl. Acad, Sci. USA, 89, 2995 (1979). T. Lazaridis and M. Karplus, J. Chem. Phys., 105, 4294 (1996). T. S. Collett, Chem. Eng. News, 75, 60 (1997). T. P. Silverstein, J. Chem. Educ., 75, 116 (1998). S. Furutaka and S. Ikawa, J. Chem. Phys., 108, 5159 (1998). M. Ikeguchi, S. Seishi, S. Nakamura, and K. Shimizu, J. Phys. Chem. B, 102, 5891 (1998). Y. Marcus, J. Solution Chem., 21, 1217 (1992). D. V. Matyushov and R. Schmid, Ber. Bunsenges. Phys. Chem., 98, 1590 (1994). V. T. Lam, P. Picker, D. Patterson, and P. Tancrede, J. Chem. Soc., Faraday Trans. 2, 70, 1465, 1479 (1974). D. Patterson, J. Solution Chem., 23, 105 (1994). P. Borothorel, J. Colloid Sci., 27, 529 (1968). H. Quinones and P. Borothorel, Compt. Rend., 277, 133 (1973). R. G. Snyder, J. Chem. Phys., 47, 1316 (1967). L. Blum and W. R. Fawcett, J. Phys. Chem., 100, 10423 (1996). Y. Marcus, Ion Solvation, Wiley, New York, 1985; Chapter 6. S. I. Pekar, Investigations in the Electronic Theory of Crystals, GIFML, Moscow (in Russian). R. A. Marcus, J. Chem. Phys., 24, 966 (1956).

776

Roland Schmid

132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161

R. R. Dogonadze and A. M. Kuznetsov, Physical Chemistry. Kinetics, VINITI, Moscow, 1973. R. A. Marcus, Disc. Farad. Soc., 29, 21 (1960). N. R. Kestner, J. Logan, and J. Jortner, J. Phys. Chem., 78, 2148 (1974). T. Holstein, Phil. Mag., B37, 499 (1978). P. Siders and R. A. Marcus, J. Am. Chem. Soc., 103, 748 (1981). A. A. Ovchinnimov and M. Ya. Ovchinnikova, Sov. Phys. JETF, 56, 1278 (1969). G. E. McManis, A. Gochev, and M. J. Weaver, Chem. Phys., 152, 107 (1991). G. M. Tom, C. Creutz, and H. Taube, J. Am. Chem. Soc., 96, 7828 (1974). H. Heitele, F. Pollinger, S. Weeren, and M. E. Michel-Beyerle, Chem. Phys., 143, 325 (1990). M. J. Powers and T. J. Meyer, J. Am. Chem. Soc., 98, 6731 (1976). P. P. Levin and P. K. N. Raghavan, Chem. Phys. Lett., 171, 309 (1990). I. R. Gould and S. Farid, J. Am. Chem. Soc., 110, 7833 (1988). G. L. Closs, L. T. Calcaterra, N. J. Green, K. W. Penfield, and J. R. Miller, J. Phys. Chem., 90, 3673 (1986). J. R. Miller, J. V. Beitz, and R. K. Henderson, J. Am. Chem. Soc., 106, 5057 (1984). D. Rehm and A. Weller, Isr. J. Chem., 8, 259 (1970). P. Suppan, Topics in Current Chemistry, Vol. 163, Springer, Berlin, 1992. S. Efrima and M. Bixon, Chem. Phys. Lett., 25, 34 (1974). N. R. Kestner, J. Logan, and J. Jortner, J. Phys. Chem., 78, 2148 (1974). T. Kakitani and N. Mataga, Chem. Phys. Lett., 124, 437 (1986); J. Phys. Chem., 91, 6277 (1987). M. Tachiya, Chem. Phys. Lett., 159, 505 (1989); J. Phys. Chem., 93, 7050 (1989). F. Markel, N. S. Ferris, I. R. Gould, and A. B. Myers, J. Am. Chem. Soc., 114, 6208 (1992). K. Kulinowski, I. R. Gould, and A. B. Myers, J. Phys. Chem., 99, 9017 (1995). B. M. Britt, J. L. McHale, and D. M. Friedrich, J. Phys. Chem., 99, 6347 (1995). L. Perera and M. L. Berkowitz, J. Chem. Phys., 96, 3092 (1992). D. V. Matyushov, Mol. Phys., 79, 795 (1993). H. Heitele, P. Finckh, S. Weeren, F. Pöllinger, and M. E. Michel-Beyerle, J. Phys. Chem., 93, 5173 (1989). E. Vauthey and D. Phillips, Chem. Phys., 147, 421 (1990). H. B. Kim, N. Kitamura, Y. Kawanishi, and S. Tazuke, J. Am. Chem. Soc., 109, 2506 (1987). F. D. Lewis and B. E. Cohen, J. Phys. Chem., 98, 10591 (1994). A. M. Swinnen, M. Van der Auweraer, F. C. De Schryver, K. Nakatani, T. Okada, and N. Mataga, J. Am. Chem. Soc., 109, 321 (1987). D. V. Matyushov and R. Schmid, Chem. Phys. Lett., 220, 359 (1994). D. V. Matyushov and R. Schmid, Mol. Phys., 84, 533 (1995). D. V. Matyushov and R. Schmid, J. Phys. Chem., 98, 5152 (1994). C. M. Elliott, D. L. Derr, D. V. Matyushov, and M. D. Newton, J. Am. Chem. Soc., 120, 11714 (1998). D. L. Derr and C. M. Elliott, J. Phys. Chem. A, 103, 7888 (1999). P. Vath, M. B. Zimmt, D. V. Matyushov, and G. A. Voth, J. Phys. Chem. B, 103, 9130 (1999). M. Born, Z. Phys., 1, 45 (1920). B. E. Conway, Ionic Hydration in Chemistry and Biophysics, Elsevier, Amsterdam, 1981. J. F. Böttcher, Theory of Electric Polarization, Elsevier, Amsterdam, 1973, Vol. 1. W. M. Latimer, K. S. Pitzer, and C. M. Slansky, J. Chem. Phys., 7, 108 (1939). A. Voet, Trans. Faraday Soc., 32, 1301 (1936). A. A. Rashin and B. Honig, J. Phys. Chem., 89, 5588 (1985). R. H. Stokes, J. Am. Chem. Soc., 86, 979 (1964); J. Am. Chem. Soc., 86, 982 (1964). K. J. Laidler and C. Pegis, Proc. R. Soc., A241, 80 (1957). B. E. Conway and E. Ayranci, J. Solution Chem., 28, 163 (1999). B. Roux, H. A. Yu, and M. Karplus, J. Phys. Chem., 94, 4683 (1990). M. Mezei and D. L. Beveridge, J. Chem. Phys., 74, 6902 (1981). Y. Marcus, J. Solution Chem., 12, 271 (1983). Y. Marcus, Chem. Rev., 88, 1475 (1988). D. Y. C. Chan, D. J. Mitchell and B. W. Ninham, J. Chem. Phys., 70, 2946 (1979). L. Blum and W. R. Fawcett, J. Phys. Chem., 96, 408 (1992). W. R. Fawcett, J. Phys. Chem., 97, 9540 (1993). A. M. Pendás, A. Costales, and V. Luana, J. Phys. Chem. B, 102, 6937 (1998). M. O’Keeffe and B. G. Hyde, Structure and Bonding in Crystals, Academic Press, New York, 1981, vol. 1, p. 227. W. H. Baur, Cryst. Rev., 1, 59 (1987). B. S. Gourary and F. J. Adrian, Solid State Phys., 10, 127 (1960).

162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187

13.2 Solvent effects on free radical polymerization

777

S. W. Rick and B. J. Berne, J. Am. Chem. Soc., 116, 3949 (1994). J. K. Hyun and T. Ichiye, J. Chem. Phys., 109, 1074 (1998). J. K. Hyun and T. Ichiye, J. Phys. Chem. B, 101, 3596 (1997). D. V. Matyushov and B. M. Ladanyi, J. Chem. Phys., 110, 994 (1999). A. Papazyan and A. Warshel, J. Chem. Phys., 107, 7975 (1997). D. W. Smith, J. Chem. Educ., 54, 540 (1977). G. E. McManis, A. Gochev, R. M. Nielson, and M. J. Weaver, J. Phys. Chem., 93, 7733 (1989). I. Rips, J. Klafter, and J. Jortner, J. Chem. Phys., 88, 3246 (1988). W. R. Fawcett and L. Blum, Chem. Phys. Lett., 187, 173 (1991). H. Heitele, P. Finckh, S. Weeren, F. Pollinger, and M. E. Michel-Beyerle, J. Phys. Chem., 93, 325 (1990). Quantitative Treatments of Solute/Solvent Interactions, P. Politzer and J. S. Murray, Eds., Elsevier, Amsterdam, 1994. 199 A. A. Gorman, M. G. Hutchings, and P. D. Wood, J. Am. Chem. Soc., 118, 8497 (1996).

188 189 190 191 192 193 194 195 196 197 198

13.2 SOLVENT EFFECTS ON FREE RADICAL POLYMERIZATION Michelle L. Coote and Thomas P. Davis Centre for Advanced Macromolecular Design, School of Chemical Engineering & Industrial Chemistry, The University of New South Wales, Sydney, Australia

13.2.1 INTRODUCTION Free radical polymerization is one of the most useful and lucrative fields of chemistry ever discovered - recent years have seen a tremendous increase in research into this area once considered a mature technological field. Free radical synthetic polymer chemistry is tolerant of diverse functionality and can be performed in a wide range of media. Emulsion and suspension polymerizations have been established as important industrial processes for many years. More recently, the ‘green’ synthesis of polymers has diversified from aqueous media to supercritical fluids and the fluorous biphase. An enduring feature of the research literature on free radical polymerization has been studies into specific solvent effects. In many cases the influence of solvent is small, however, it is becoming increasingly evident that solvent effects can be used to assist in controlling the polymerization reaction, both at the macroscopic and at the molecular levels. The purpose of this chapter is to give a brief introduction to the types of specific solvent effect that can be achieved in both free radical homo- and co-polymerizations. 13.2.2 HOMOPOLYMERIZATION Free radical polymerization can be conveniently codified according to the classical chain reaction steps of initiation, propagation, transfer and termination. In cases where a significant solvent effect is operative then the effect is normally exerted in all of these steps. However, for the purpose of facilitating discussion this chapter is broken down into these specific reaction steps. 13.2.2.1 Initiation Solvent effects on the initiation reaction are primarily on the rate of decomposition of initiator molecules into radicals and in the efficiency factor, f, for polymerization. However, in some instances the solvent plays a significant role in the initiation process, for example, in

778

Michelle L. Coote and Thomas P. Davis

initiation reactions with t-butoxy radical where the primary radical rarely initiates a chain but instead abstracts a hydrogen atom from the solvent medium, which subsequently initiates the chain.1 The consequence of this is that the polymer chains contain fragments of solvent. As the stability of the chains to thermal and photochemical degradation is governed, in part, by the nature of the chain ends then the solvent moieties within the chain can have a substantial impact on the material performance of the polymer. The efficiency factor, f, decreases as the viscosity of the reaction medium increases.2 This is caused by an increase in the radical lifetime within the solvent cage, leading to an increased possibility of radical-radical termination. In this regard the diffusion rates of the small radicals becomes an important consideration and Terazima and co-workers3,4 have published results indicating that many small radicals diffuse slower than expected. They have attributed this to specific interactions between radical and solvent molecules. 13.2.2.2 Propagation The ability of solvents to affect the homopropagation rate of many common monomers has been widely documented. For example, Bamford and Brumby5 showed that the propagation rate (kp) of methyl methacrylate (MMA) at 25°C was sensitive to a range of aromatic solvents. Burnett et al.6 found that the kp of styrene (STY) was depressed by increasing concentrations of benzonitrile, bromobenzene, diethyl phthalate, dinonyl phthalate and diethyl malonate, while in other studies7,8 they found that the kp for MMA was enhanced by halobenzenes and naphthalene. More recent work by Zammit et al.9 has shown that solvents capable of hydrogen-bonding, such as, benzyl alcohol and N-methyl pyrrolidone have a small influence on both the activation energy (Ea) and pre-exponential factor (A) in STY and MMA homopropagation reactions. These are but a few of the many instances of solvent effects in the homopolymerization reactions of two typical monomers, STY and MMA. For these monomers, solvent effects are relatively small, and this is indicative of the majority of homopropagation reactions. However, in some instances much larger effects are observed, especially in cases where specific interactions such as H-bonding or ionization occur. Examples of this type include the polymerization of N-vinyl-2-pyrrolidone (where water has been found to dramatically increase kp)10 and the polymerization of acrylamide (where pH plays a strong role).11 There is only limited data on the Arrhenius parameters for homopropagation reactions in different solvents and this indicates that both the activation energy and pre-exponential factor are affected.9,12 In some cases the solvent effect is not on the elementary rate constant kp but on the local monomer concentration (sometimes referred to as the ‘Bootstrap" effect). This effect can originate in the preferential solvation of either the monomer (which is always present as a solvent) or the added solvent. It has also been suggested that in some instances the growing polymer coil can ‘shield’ the radical chain-end resulting in a low monomer concentration. This shielding effect would be expected to be greatest in poor solvents (hence a tighter coil).13 For methyl methacrylate and styrene the largest solvent effects on propagation seem to be in the order of a 40% change in kp.14,15 In some solvents there seems to be reasonably strong evidence that the solvent does cause changes to the geometry of the transition state (e.g., dimethyl formamide and acetonitrile in styrene polymerization)14 and in liquid carbon dioxide it appears that the 40% change in kp for methyl methacrylate can be ascribed to the poor solvent medium.16 Recent work has found that some fluoro-alcohols17 can influence the tacticity of free radical polymerization lending further credence to the concept of solvent-induced changes to the transition state of


779

the radical addition reaction. The largest solvent effects observed on kp for homopropagations have been for vinyl acetate18 and for α-(hydroxymethyl) ethyl acrylate.12 In the former case the radical is highly unstable and some form of π-complexation between the vinyl acetate radical and aromatic solvents seems plausible. However, the large solvent effect cannot be explained by a simple radical stabilization argument (because of the early transition state for free radical propagation reactions)19 and again the evidence points towards a change in the geometry of the transition state. The solvent effects on α-(hydroxymethyl) ethyl acrylate are in the order of 300% on kp and there are large changes in both the Arrhenius parameters as the solvent medium is changed.12 In monomers exhibiting a strong solvent effect on propagation it is plausible that some control of the stereochemistry of the chains is possible by manipulating the solvent and possibly utilizing Lewis acids as additives. This approach is already being successfully applied to the control of radical reactions in conventional organic chemistry.20 13.2.2.3 Transfer Solvent effects on transfer reactions have not received too much attention. It would be expected (owing to the similarities between the transition states for radical addition and abstraction reactions) that these solvent effects should emulate those found in propagation reactions. However, there is potential for significant polar interactions in transfer reactions. Odian21 has suggested that polar interactions play a significant role in the transfer reactions between styrene and carbon tetrachloride. More recent work supports this idea.22 Significant solvent effects have been observed in catalytic chain transfer reactions using cobaloximes where the transfer reaction appears (in some cases at least) to be diffusion controlled and therefore the speed of the reaction is governed, in part, by the viscosity of the polymerizing medium.23 In transfer reactions involving organometallic reagents then solvent effects may become important where ligand displacement may occur. This is thought to happen in catalytic chain transfer when pyridine is utilized as a solvent.24 13.2.2.4 Termination The solvent effects on the termination reaction have been extensively studied. In early work, it was established that the radical-radical termination reaction is diffusion controlled and the efficacy of termination was found to have a strong relationship with the solvent viscosity.25 Subsequently, more complex models have been developed accounting for the quality of the solvent (hence the size of the polymer coil).26 The current debate centers on the relative roles played by segmental and translational diffusion at different stages of conversion for a variety monomers. Clearly in both cases the nature of the solvent becomes important. Solvent effects are known to play a significant role in determining the strength and onset conversion of the gel effect. This work originated in the classical paper by Norrish and Smith27 who reported that poor solvents cause an earlier gel effect in methyl methacrylate polymerization. Recent careful studies of the gel effect by Torkelson and co-workers28 have reinforced observations made by Cameron and Cameron29 over two decades ago concluding that termination is hindered in poor solvents due to formation of more tightly coiled polymer radicals. 13.2.3 COPOLYMERIZATION When solvent effects on the propagation step occur in free-radical copolymerization reactions, they result not only in deviations from the expected overall propagation rate, but also in deviations from the expected copolymer composition and microstructure. This may be

780


true even in bulk copolymerization, if either of the monomers exerts a direct effect or if strong cosolvency behavior causes preferential solvation. A number of models have been proposed to describe the effect of solvents on the composition, microstructure and propagation rate of copolymerization. In deriving each of these models, an appropriate base model for copolymerization kinetics is selected (such as the terminal model or the implicit or explicit penultimate models), and a mechanism by which the solvent influences the propagation step is assumed. The main mechanisms by which the solvent (which may be one or both of the comonomers) can affect the propagation kinetics of free-radical copolymerization reactions are as follows: (1) Polarity effect (2) Radical-solvent complexes (3) Monomer-solvent complexes (4) Bootstrap effect In this chapter we explain the origin of these effects, show how copolymerization models for these different effects may be derived, and review the main experimental evidence for and against these models. Throughout this review the baseline model for copolymerization is taken as the terminal or Mayo-Lewis model.30 This model can be used to derive well-known expressions for copolymer composition and copolymerization propagation kinetics. Deviations from this model have often been interpreted in terms of either solvent effects or penultimate unit effects, although the two are by no means mutually exclusive. Deviations which affect both the copolymer composition and propagation kinetics have been termed explicit effects by Fukuda31 in deriving penultimate unit models, whereas deviations from the kinetics without influencing the copolymer composition have been termed implicit effects. In this review we use the same terminology with respect to solvent effects: that is, a solvent effect on kp only is termed an implicit solvent effect, while a solvent effect on composition, microstructure and kp is termed explicit. The relatively recent discovery by Fukuda and co-workers32 of the seemingly general failure of the terminal model to predict kp, even for bulk copolymerizations that follow the terminal model composition equation, led them to propose an implicit penultimate unit effect as a general phenomenon in free-radical copolymerization kinetics. We conclude this review with a brief examination of the possibility that a implicit solvent effect, and not an implicit penultimate unit effect, may instead be responsible for this failure of the terminal model kp equation. 13.2.3.1 Polarity effect 13.2.3.1.1 Basic mechanism One type of solvent effect on free-radical addition reactions such as the propagation step of free-radical polymerization is the so-called ‘polarity effect’. This type of solvent effect is distinguished from other solvent effects, such as complexation, in that the solvent affects the reactivity of the different types of propagation steps without directly participating in the reaction. The mechanism by which this could occur may be explained as follows. The transition states of the different types of propagation steps in a free-radical copolymerization may be stabilized by charge transfer between the reacting species. The amount of charge transfer, and hence the amount of stabilization, is inversely proportional to the energy difference between the charge transfer configuration, and the product and reactant configurations that combine to make up the wave function at the transition state.33 Clearly, the stability of the charge transfer configuration would differ between the cross- and


781

homopropagation reactions, especially in copolymerization of highly electrophilic and nucleophilic monomer pairs. Hence, when it is significant, charge transfer stabilization of the transition state occurs to different extent in the cross- and homopropagation reactions, and thus exerts some net effect on the monomer reactivity ratios. Now, it is known that polar solvents can stabilize charged species, as seen in the favorable effect of polar solvents on both the thermodynamics and kinetics of reactions in which charge is generated.34 Therefore, when charge transfer in the transition state is significant, the stability of the charge transfer species and thus the transition state would be affected by the polarity of the solvent, and thus a solvent effect on reactivity ratios would result. 13.2.3.1.2 Copolymerization model There are two cases to consider when predicting the effect of solvent polarity on copolymerization propagation kinetics: (1) the solvent polarity is dominated by an added solvent and polarity is thus independent of the comonomer feed ratio, or (2) the solvent polarity does depend on the comonomer feed ratio, as it would in a bulk copolymerization. In the first case, the effect on copolymerization kinetics is simple. The monomer reactivity ratios (and additional reactivity ratios, depending on which copolymerization model is appropriate for that system) would vary from solvent to solvent, but, for a given copolymerization system they would be constant as a function of the monomer feed ratios. Assuming of course that there were no additional types of solvent effect present, these copolymerization systems could be described by their appropriate base model (such as the terminal model or the explicit or implicit penultimate models), depending on the chemical structure of the monomers. In the second case, the effect of the solvent on copolymerization kinetics is much more complicated. Since the polarity of the reacting medium would vary as a function of the comonomer feed ratios, the monomer reactivity ratios would no longer be constant for a given copolymerization system. To model such behavior, it would be first necessary to select an appropriate base model for the copolymerization, depending on the chemical structure of the monomers. It would then be necessary to replace the constant reactivity ratios in this model by functions of the composition of the comonomer mixture. These functions would need to relate the reactivity ratios to the solvent polarity, and then the solvent polarity to the comonomer feed composition. The overall copolymerization kinetics would therefore be very complicated, and it is difficult to suggest a general kinetic model to describe these systems. However, it is obvious that such solvent effects would cause deviations from the behavior predicted by their appropriate base model and might therefore account for the deviation of some copolymerization systems from the terminal model composition equation. 13.2.3.1.3 Evidence for polarity effects in propagation reactions The idea of charge separation in the transition state of the propagation step of free radical polymerization reactions, as suggested by Price,35 was discounted by Mayo and Walling36 and many subsequent workers.37 Their rejection of this idea was based upon the absence of any unambiguous correlation between the reactivity ratios of a system and the dielectric constant of the solvent. For instance, in the copolymerization of STY with MMA, it was reported that the reactivity ratios were independent of small quantities of water, ethyl benzene, dodecylmercaptans or hydroquinone, or the presence or absence of air30,38,39 and were thus unaffected by the dielectric constant of the system. In contrast, other studies have found a relationship between dielectric constant and the reactivity ratios in specific systems.40,41

782


The apparent lack of a general relationship between the dielectric constant of the system and the monomer reactivity ratios does not necessarily discount a polarity effect on reactivity ratios. A polarity effect is only expected to occur if the charge transfer configurations of the transition state are sufficiently low in energy to contribute to the ground state wave function. Since this is not likely to occur generally, a comprehensive correlation between reactivity ratios and the solvent dielectric constant is unlikely. Furthermore, even in systems for which a polarity effect is operating, a correlation between solvent dielectric constant and monomer reactivity ratios may be obscured by any of the following causes. • The operation of additional types of solvent effect, such as a Bootstrap effect, that would complicate the relationship between solvent polarity and reactivity ratios. • Errors in the experimental data base from which the correlation was sought. • The recognized inadequacy of simple reactivity - dielectric constant correlations, that take no account of specific interactions between the solvent and solute molecules.34 In fact, recent theoretical33 and experimental studies42 of small radical addition reactions indicate that charge separation does occur in the transition state when highly electrophilic and nucleophilic species are involved. It is also known that copolymerization of electron donor-acceptor monomer pairs are solvent sensitive, although this solvent effect has in the past been attributed to other causes, such as a Bootstrap effect (see Section 13.2.3.4). Examples of this type include the copolymerization of styrene with maleic anhydride43 and with acrylonitrile.44 Hence, in these systems, the variation in reactivity ratios with the solvent may (at least in part) be caused by the variation of the polarity of the solvent. In any case, this type of solvent effect cannot be discounted, and should thus be considered when analyzing the copolymerization data of systems involving strongly electrophilic and nucleophilic monomer pairs. 13.2.3.2 Radical-solvent complexes 13.2.3.2.1 Basic mechanism Solvents can also interfere in the propagation step via the formation of radical-solvent complexes. When complexation occurs, the complexed radicals are more stable than their corresponding uncomplexed-radicals, as it is this stabilization that drives the complexation reaction. Thus, in general, one might expect complexed radicals to propagate more slowly than their corresponding free-radicals, if indeed they propagate at all. However, in the special case that one of the comonomers is the complexing agent, the propagation rate of the complexed radical may instead be enhanced, if propagation through the complex offers an alternative less-energetic reaction pathway. In any case, the complexed radicals would be expected to propagate at a rate different to their corresponding free-radicals, and thus the formation of radical-solvent complexes would affect the copolymerization propagation kinetics. 13.2.3.2.2 Copolymerization model A terminal radical-complex model for copolymerization was formulated by Kamachi.37 He proposed that a complex is formed between the propagating radical chain and the solvent (which may be the monomer) and that this complexed radical has a different propagation rate constant to the equivalent uncomplexed radical. Under these conditions there are eight different propagation reactions in a binary copolymerization, assuming that the terminal unit is the only unit of the chain affecting the radical reactivity. These are as follows.


783

k

ij RMi ⋅ +Mj  → RMiMj ⋅ or RMiMj ⋅ S where: i, j = 1or 2

k

cij → RMiMj ⋅ or RMiMj ⋅ S where: i, j = 1or 2 RMi ⋅ S + Mj  

There are also two equilibrium reactions for the formation of the complex: Ki RMi ⋅ +S ← → RMi ⋅ S where: i, j = 1or 2

Applying the quasi-steady-state and long-chain assumptions to the above reactions, Kamachi derived expressions for ri and kii, which are used in place of ri and kii in the terminal model equations for composition and kp: k ii = k ii

1 + s ci K i [Ci ] 1 + K i [Ci ]

and

ri = ri

1 + s ci K i [Ci ]

1 + (ri / rci )s ci K i [Ci ]

where: ri = kii/kij; ric = kcii/kcij; sci = kcii/kii; i, j = 1 or 2 and i ≠ j

Variants of this model may be derived by assuming an alternative basis model (such as the implicit or explicit penultimate models) or by making further assumptions as to nature of the complexation reaction or the behavior of the complexed radical. For instance, in the special case that the complexed radicals do not propagate (that is, sci = 0 for all i), the reactivity ratios are not affected (that is, ri = ri for all i) and the complex formation serves only removal of radicals (and monomer, if monomer is the complexing agent) from the reaction, resulting in a solvent effect that is analogous to a Bootstrap effect (see Section 13.2.3.4). 13.2.3.2.3 Experimental evidence There is certainly strong experimental evidence for the existence of radical-solvent complexes. For instance, Russell45-47 and co-workers collected experimental evidence for radical-complex formation in studies of the photochlorination of 2,3-dimethylbutane in various solvents. In this work, different products were obtained in aliphatic and aromatic solvents, and this was attributed to formation of a π-complex between the Cl atom and the aromatic solvent. Complex formation was confirmed by flash photolysis.48-51 Complex formation was also proposed to explain experimental results for the addition of trichloromethane radical to 3-phenylpropene and to 4-phenyl-1-butene52 and for hydrogen abstraction of the t-butoxy radical from 2,3-dimethylbutane.53 Furthermore, complexes between nitroxide radicals and a large number of aromatic solvents have been detected.54-57 Evidence for complexes between polymer radicals and solvent molecules was collected by Hatada et al.,58 in an analysis of initiator fragments from the polymerization of MMA-d with AIBN and BPO initiators. They discovered that the ratio of disproportionation to combination depended on the solvent, and interpreted this as evidence for the formation of a polymer radical-solvent complex that suppresses the disproportionation reaction. There is also experimental evidence for the influence of radical-solvent complexes in small radical addition reactions. For instance, Busfield and co-workers59-61 used radical-solvent to explain solvent effects in reactions involving small radicals, such as t-butoxyl radicals towards various electron donor-electron acceptor monomer pairs. The observed solvent effects were interpreted in terms of complex formation between the t-butoxyl radical and the electron-acceptor monomer, possibly via a sharing of the lone pair on the t-butoxyl oxy-

784


gen with the π-system of the acceptor monomer. Several workers have invoked frontier orbital theory to rationalize such solvent effects in terms of radical-solvent complex formation, and thus provide a theoretical base.37,62 Many workers have suggested radical-solvent complexes as an explanation for the influence of aromatic compounds on the homopolymerization of vinyl monomers. For instance, Mayo63 found that bromobenzene acts as a chain transfer agent in the polymerization of STY but is not incorporated into the polymer. He concluded that a complex is formed between the solvent molecule and either the propagating polystyryl radical or a proton derived from it. The influence of halobenzenes on the rate of polymerization of MMA was detected by Burnett et al.8,64,65 They proposed that the efficiency of a number of different initiators increased in various halogenated aromatic solvents and, since enhanced initiator or solvent incorporation into the polymer was not observed, they concluded that initiator-solvent-monomer complex participation affected the initiator efficiency. Henrici-Olive and Olive66-71 suggested that this mechanism was inadequate when the degree of polymerization was taken into account and they proposed instead a charge transfer complex between the polymer radical and aromatic solvent. The polymer radical can form a complex with either the monomer or solvent molecule, but only the former can propagate. Bamford and Brumby,5 and later Burnett et al.,72,73 interpreted their solvent-effects data for kp in terms of this donor-acceptor complex formation between aromatic solvents and propagating radicals. Radical-solvent complexes are expected to be favored in systems containing unstable radical intermediates (such as vinyl acetate) where complexation may lead to stabilization. In this regard Kamachi et al18 have noted that solvent effects on vinyl acetate homopolymerization result in a reduced kp. Kamachi et al.74 also measured the absolute rate constants of vinyl benzoate in various aromatic solvents and found that kp increased in the order: benzonitrile < ethyl benzoate < anisole < chlorobenzene < benzene < fluorobenzene < ethyl acetate

They argued that this trend could not be explained by copolymerization through the solvent or transfer to the solvent because there was no correlation with the solvent dielectric constant or polarity, or with the rate constants for transfer to solvent. However, there was a correlation with the calculated delocalization stabilization energy for complexes between the radical and the solvent, which suggested that the propagating radical was stabilized by the solvent or monomer, but the solvent did not actually participate in the reaction. As noted in the introduction to this section, radical-solvent complexes may enhance the propagation rate if propagation through the complex offers an alternative, less-energetic pathway for propagation. An example of this behavior is found in the homopolymerization of acrylamide. The homopropagation rate coefficient for this monomer shows a negative temperature dependence, which has been explained in terms of radical-complex formation. Pascal et al.11,75 suggested that propagation proceeds via a complex that enhances the propagation rate, and this complex dissociates as temperature increases, thus explaining the normal temperature dependence of the propagation rate at high temperatures. This interpretation was supported by the observation that acrylamide behaves normally in the presence of reagents such as propionamide, which would be expected to inhibit complex formation. Given the experimental evidence for the existence of radical-solvent complexes and their influence on free-radical addition reactions such as homopropagation, it is likely that


785

radical-solvent complexes will affect the copolymerization kinetics for certain copolymerization systems, and indeed many workers have invoked the radical-complex model in order to explain solvent effects in copolymerization. For instance, Heublein and Heublein76 have invoked a radical complex model in combination with a partitioning idea (see Section 13.2.3.4) to explain solvent effects on the copolymerization of vinyl acetate with acrylic acid. More recently, O’Driscoll and Monteiro77 suggested that the effect of benzyl alcohol on the copolymerization of STY-MMA was best described by an RC-type model. This was supported by pulsed-laser studies9 on the homopropagation reactions where Ea values were found to be increased slightly by the presence of benzyl alcohol. Czerwinski (see for example reference78 and references cited therein) has also published a variant of the RC model and has applied his model to a range of copolymerization experimental data. In conclusion, there is a strong experimental evidence for the importance of radical-solvent complexes in a number of specific copolymerization systems, especially when there is a large disparity in the relative stabilities of the different propagating radicals. 13.2.3.3 Monomer-solvent complexes 13.2.3.3.1 Introduction A solvent may also interfere in the propagation step via complexation with the monomer. As was the case with radical-solvent complexes, complexed monomer might be expected to propagate at a different rate to free monomer, since complexation might stabilize the monomer, alter its steric properties and/or provide an alternative pathway for propagation. In examining the effect of such complexation on copolymerization kinetics, there are a number of different mechanisms to consider. In the case that the complex is formed between the comonomers, there are three alternatives: (1) the monomer-monomer complex propagates as a single unit, competing with the propagation of free monomer; (2) the monomer-monomer complex propagates as a single unit, competing with the propagation of free monomer, but the complex dissociates during the propagation step and only one of the monomers is incorporated into the growing polymer radical; (3) the monomer-monomer complex does not propagate, and complexation serves only to alter the free monomer concentrations. In the case that the complex is formed between one of the monomers and an added solvent, there are two further mechanisms to consider: (4) the complexed monomer propagates, but at a different rate to the free monomer; (5) the complexed monomer does not propagate. Models based on mechanisms (1) and (2) are known as the monomer-monomer complex participation (MCP) and dissociation (MCD) models, respectively. Mechanisms (3) and (5) would result in a solvent effect analogous to a Bootstrap effect, and will be discussed in Section 13.2.3.4. In this section, we review the MCP and MCD models, and conclude with a brief discussion of specific monomer-solvent interactions. 13.2.3.3.2 Monomer-monomer complex participation model The use of monomer-monomer charge transfer complexes to explain deviations from the terminal model was first suggested by Bartlet and Nozaki,79 later developed by Seiner and Litt,80 and refined by Cais et al.81 It was proposed that two monomers can form a 1:1 donor complex and add to the propagating chain as a single unit in either direction. The complex would be more reactive because it would have a higher polarizability due to its larger π-electron system that can interact more readily with the incoming radical. The complex would also have a higher pre-exponential factor, as a successful attack may be achieved over a

786


wider solid angle.80 The heavier mass of the complex would also serve to increase the pre-exponential factor. In addition to the four terminal model reactions, four complex addition reactions and an equilibrium constant are required to describe the system. k

ij → RMiMj ⋅ where: i, j = 1or 2 RMi ⋅ +Mj o 

k

iij → RMiMiMj ⋅ where: i, j = 1or 2 and i ≠ j RMi ⋅ +MiMj 

k

iji → RMiMjMi ⋅ where: i, j = 1or 2 and i ≠ j RMi ⋅ +MiMj 

K → M1M 2 M1o + M 2 o ←

The composition and propagation rate can be expressed in terms of the following parameters.31 o F1 f1 (A2 B1 )r1f1 + (A1C2 )f 2 = o F2 f 2 (A1B 2 )r2 f 2 o + (A2 C1 )f1 o o

kp =

(A2 B1 )r1 (f 2 o )

2

o

( ) + (A C + A C )f ) + (A r f / k )

+ (A1B 2 )r2 f 2

(A r f

2 11

o

/ k11

o

2

1

2

2

1

1

o

f2

o

o

1 2 2

22

where: A1 = 1 + r1s1cQf1q and A2 = 1 + r2s2cQf2q B1 = 1 + s1c(1 + r1c-1)Qf2q and B2 = 1 + s2c(1 + r2c-1)Qf1q C1 = 1 + r1s1c(1 + r1c-1)Qf1q and C2 = 1+ r2s2c(1 + r2c-1)Qf2q 2Qfiq = {[Q(fj - fi) +1]2 +4Qfi}1/2 - [Q(fi - fj) +1] and Q = k[M] fi is feed composition of Mi fiq = [Miq]/[M] ri = kii/kij; ric = kiij/kiji; sic = kiij/kii where: i,j = 1 or 2 and i ≠ j

The applicability of the MCP model to strongly alternating copolymerization has been a long standing point of contention. In essence, there are two opposing accounts of the strongly alternating behavior observed in copolymers of electron-donor-acceptor (EDA) monomer pairs. In the first account, this behavior has been attributed to the fact that the transition state is stabilized in cross-propagation reaction and destabilized in the homopropagation. Deviations from the terminal model are caused merely by penultimate unit effects. In the second account -the MCP model- the strongly alternating behavior is a result of propagation of a 1:1 comonomer complex which, as seen above, also leads to deviations from the terminal model. An intermediate mechanism, which will be discussed shortly, is the MCD model in which the complex dissociates during the propagation step. The main approach to discriminating between these models has been to compare their ability to describe the copolymerization data of various explicit systems, and to study the effect of added solvents on their behavior. Unfortunately, both approaches have led to inconclusive results.


787

As an example, the system STY with maleic anhydride (MAH) has been perhaps the most widely studied EDA system and yet there is still uncertainty concerning the role of the EDA complex in its propagation mechanism. Early studies82,83 concluded that its behavior was best modelled by a penultimate model, despite the spectroscopic evidence for EDA complexes in this system. Later Tsuchida et al.84,85 fitted an MCP model, based on the evidence that the rate went through a maximum at 1:1 feed ratio in benzene or CCl4 but in strong donor solvents no such maximum occurred and instead the rate increased with the content of MAH in the feed. They argued that maximum in rate at 1:1 feed ratios was due to the fact that propagation occurred via the complex, which had a maximum concentration at this point. In strong donor solvents, the maximum rate moved to higher concentrations of MAH due to competition between the donor and STY for complexation with the MAH. However, a few years later, Dodgson and Ebdon86,87 conducted an extensive study of STY-MAH in various solvents and discounted the MCP model on the basis of an absence of a dilution effect with the inert solvent MEK. In an MCP model a dilution effect would be expected due to the decrease in the relative concentration of the comonomer complex and the enhanced participation of the free monomer.88 Later, Farmer et al.89 reanalyzed this data and concluded that the composition data was consistent with both models and suggested sequence distribution may provide the answer. They also pointed out that there was a small dilution effect in MEK -greater than that predicted by the penultimate model and less than that predicted by the MCP model. Hill et al.90 has suggested that interpretation of the effect of solvents is complicated by the fact that no solvent is truly inert, hence such results such be treated with caution. More recently Sanayei et al.91 have performed a pulsed-laser polymerization study on STY-MAH copolymerization in butanone and acetonitrile. They concluded that whilst the complex participation model described the copolymer composition it failed to predict the average kp data. Consequently the best description of this copolymerization was given by the penultimate unit model. There have been many other systems for which the MCP model has been proposed as an alternative to the penultimate unit model. For instance, Litt and Seiner used the MCP model to describe the composition of a number of systems, including MAH with 1-diphenylethylene, β-cyanacrolein with styrene,92 and vinyl acetate with dichlorotetrafluoroacetone and with hexafluoroacetone.80 An MCP model has also been suggested for the system STY-SO2.39,83,93-95 In this system, the composition changes with dilution or with solvent changes, strongly alternating behavior is observed across a range of feed ratios, and one of the comonomers (SO2) does not undergo homopolymerization. However, while the MCP model appears to be appropriate for some systems, in other strongly alternating copolymerizations it is clearly not appropriate. For instance, there are many strongly alternating copolymerizations for which there is no evidence of complex formation.36,39,88,96 Even when complex formation is known to occur, results cannot always be explained by the MCP model. For instance, measurements of sequence distribution data revealed that, while both the MCP and penultimate model could provide an adequate description of the composition of STY with acrylonitrile (AN), only the penultimate model could account for the sequence distribution data for this system.97 As will be seen shortly, there is evidence that in some systems the heat of propagation would be sufficient to dissociate the EDA complex and hence it could not add to the monomer as unit. In this case an MCD model would be more appropriate. Thus, it might be concluded that the MCP, MCD and the penultimate models are needed to describe the behavior of strongly alternating systems, and

788


the selection of each model should be on a case-by-case basis. There have been many more systems for which the MCP model has been evaluated against the penultimate model on the basis of kinetic behavior. These studies have been extensively reviewed by Hill et al.90 and Cowie98 will therefore not be reviewed here. Instead a few additional sources of evidence for the participation of the EDA complex will be highlighted. UV and NMR evidence for the existence of EDA complexes There is certainly a demonstrable existence of comonomer complexes in solutions of electron donor acceptor monomer pairs. These complexes can be detected, and their equilibrium constants measured, using UV or NMR spectroscopy. Techniques for this are described in detail in reviews of comonomer complexes by Cowie98 and Hill et al.90 The latter review90 also includes a listing of the equilibrium constants for the numerous EDA complexes that have been experimentally detected. The existence of comonomer complexes is not sufficient evidence for their participation in the propagation step of copolymerization, but the fact that they exist in solutions from which strongly alternating copolymers are produced suggests that they play some role in the mechanism. Furthermore, the ability to measure their strengths and quantify the effects of solvents on their observed equilibrium constants without performing kinetic experiments, may provide the key to establishing their role in the propagation mechanism. Since the alternative models for copolymerization include (or in some cases omit) the equilibrium constant for these complexes in different ways; if the equilibrium constant was to be measured separately and not treated as an adjustable parameter in the kinetic analysis, more sensitive model discrimination would be possible. To date, such an analysis does not appear to have been performed but it should be included in subsequent kinetic analyses of these explicit systems. Temperature effects The study of the temperature dependence of copolymerization behavior may also provide evidence for the role of comonomer complexes. As was seen previously in the study of acrylamide, complexes dissociate at high temperatures and hence, if the complex is involved in controlling an aspect of the polymerization behavior, a change in this behavior should be observed at the temperature corresponding to the complete dissociation of the complex. Such evidence has been obtained by Seymour and Garner99,100 for the copolymerization of MAH with a variety of vinyl monomers, including STY, VA, AN, and α-MSTY. They observed that the copolymers undergo a change from strongly alternating to random at high temperatures, and these temperatures are also the temperatures at which the concentration of the EDA complex becomes vanishingly small. It is true that, since reactivity ratios have an enthalpy component, they approach unity as temperature increases. Hence, most models predict that the tendency of copolymers to form a random microstructure increases as temperature increased. Indeed, more recent work by Klumperman101 has shown that for STY-MAH copolymerization, the reactivity ratios do follow an Arrhenius type of temperature dependence. However, further work is required to verify this for the other copolymerization listed above. Based upon the existing copolymerization data, it appears that for many systems there are sudden transition temperatures that correspond to the dissociation of the complex, which does suggests that the complex is in some way responsible for the alternating behavior.99,100 Stereochemical evidence for the participation of the complex Stereochemical data may provide evidence for participation of the EDA complex. The EDA complex will prefer a certain geometry -that conformation in which there is maximum


789

overlap between the highest occupied molecular orbital (HOMO) of the donor and the lowest unoccupied molecular orbital (LUMO) of the acceptor.102 If the complex adds to the propagating polymer chain as a unit, then this stereochemistry would be preserved in the polymer chain. If, however, only free monomer addition occurs, then the stereochemistry of the chain should be completely random (assuming of course that there are no penultimate unit effects operating). Hence, it is possible to test for the participation of the monomer complexes in the addition reaction by examination of the stereochemistry of the resulting polymer. Such stereochemical evidence has been collected by a number of workers. For instance, Iwatsuki and Yamashia103 observed an unusually high percentage of cis units in MAH/butadiene copolymers. Olson and Butler104 studied the EDA system N-phenylmaleimide (NPM)/2-chloroethyl vinyl ether (CEVE) and found that the stereochemistry at succinimide units in NPM-CEVE copolymers is predominantly cis, and random elsewhere. Furthermore, they noted that the proportion of cis units was correlated with those variables with which the concentration of the EDA complex was also correlated. In these examples, the cis geometry is that which is most stable for the complex. However, Rätzsch and Steinert105 have argued that this preference for cis geometry may also be explained by propagation occurring via a complex between the reacting free monomer and the chain end, as in an RC model. Thus this evidence should be used in conjunction with other evidence for model discrimination. Further stereochemical evidence for the MCP model has been obtained by Butler et al.106 They predicted that the usual preference for head-tail addition in free-radical polymerization would be overcome if propagation occurred via the EDA complex, and its favored geometry was a head-head conformation. They noted that for most EDA pairs head-tail geometry was favored and hence the predominance of head-tail linkages in these copolymers could not discriminate between free monomer addition and complex participation. To solve this problem, they designed and synthesized two monomer pairs for which a head-head conformation would be expected in their EDA complexes. These were the systems dimethyl cyanoethylene dicarboxylate (DMCE) with CEVE, and dimethyl cyanoethylene dicarboxylate (DMCE) with CEVE. They then showed that there were significant head-head linkages in the resulting copolymer and the proportion of these linkages was correlated with same types of variables that had previously affected the cis content of NPM/CEVE copolymers -that is, those variables which affected the concentration of the EDA complex. Thus they concluded that there was strong stereochemical evidence for the participation of the EDA complex in the propagation step. ESR evidence for the participation of the complex ESR studies have also been suggested as a means for providing information about the participation of the EDA complex. Since the addition of the complex is likely to occur more readily in one direction, if propagation occurs as the repeated addition of the complex then the propagating radical should be predominantly of one type. However, if free monomer addition predominantly occurs, both types of radical are likely to be present at any time. Thus ESR can be used to distinguish between the two mechanisms. This approach was used by Smirnov et al.107 to show that, in the system phenyl vinyl ether/MAH, alternating addition of the free monomer predominates, but participation of EDA complexes is important for the system butyl vinyl ether/MAH. They argued that the difference in the behavior of the two EDA systems was a result of the different strengths of their EDA complexes. In another

790


study, Golubev et al.,108 used ESR to show that for dimethylbutadiene/MAH the cross-propagation of the free monomers dominated. However, Barton et al.109 has questioned the assignments of ESR signals in the previous studies and suggested that the ESR evidence was inconclusive. Furthermore, the predominance of one type of ESR signal may also be explained without invoking the MCP model. Assuming that cross-propagation is the dominant reaction, and that one of the radicals is much less stable than the other, it might reasonably be expected that the less stable radical would undergo fast cross-propagation into the more stable radical, resulting in an ESR signal dominated by the more stable radical. Hence it appears that ESR is not able to discriminate between this and the MCP mechanism. 13.2.3.3.3 Monomer-monomer complex dissociation model Tsuchida and Tomono84 suggested that the monomer-monomer complex described in the MCP model may dissociate upon addition to the chain, with only one unit adding. The concept was developed by Karad and Schneider110 who argued that the dissociation of the complex is likely since its heat of formation is typically less than the heat of propagation. As an example, they measured the heat of formation for a STY/fumaronitrile complex, and found that it was only 1.6 kcal/mol, significantly less than the heat of propagation (15-20 kcal/mol). Under a complex-dissociation mechanism, the role of the complex is merely to modify the reactivity of the reactant monomers. A model based on the complex-dissociation mechanism was first formulated by Karad and Schneider110 and generalized by Hill et al.111 Again, eight rate constants and two equilibrium constants are required to describe the system. k

ij → RMiMj ⋅ where: i, j = 1or 2 RMi ⋅ +Mj 

k

ijc → RMiMj ⋅ where: i, j = 1or 2 RMi ⋅ +MjC  

Ki → MiC Mi + C ←

where: i, j = 1or 2

As for the previous models, expressions for kp and composition can be derived in terms of these parameters by first calculating kii and ri and then using them in place of kii and ri in the terminal model equations. The relevant formulae are: k ii = k ii

1 + s ic K i [Ci ] 1 + K i [Ci ]

and ri = ri

1 + s ic K i [Ci ]

1 + (ri / rci )s ic K i [Ci ]

where: ri = kii/kij; ric = kiic/kijc; sic = kiic/kii; i,j = 1 or 2 and i ≠ j

Efforts to compare this model with the MCP model have been hindered by the fact that similar composition curves for a given system are predicted by both models. Hill et al.111 showed that the composition data of Dodgson and Ebdon87 for STY/MAH at 60°C could be equally well described by the MCP, MCD or penultimate unit models. They suggested that sequence distribution would be a more sensitive tool for discriminating between these models. One study which lends some support to this model over the MCP model for describing this system was published by Rätzsch and Steinert.105 Using Giese’s112 ‘mercury method’ to study the addition of monomers to primary radicals, they found that in mixtures of MAH and STY, only reaction products from the addition of free monomers, and not the EDA


791

complex, to primary cyclohexyl radicals were found. Thus they concluded that the STY/MAH complex in the monomer solution is disrupted during the propagation step. It is likely that both the MCP and MCD mechanisms are valid and their validity in a specific system will depend on the relative strength of the EDA complex concerned. The MCD model may be useful for accounting for those systems, in which EDA complexes are known to be present but the MCP model has been shown not to hold. 13.2.3.3.4 Specific solvent effects Several monomers are particularly susceptible to strong solvent effects via specific interactions such as hydrogen bonding, ionic strength and pH. The kinetic consequences of these specific interactions will vary from system to system. In some cases the radical and/or monomer reactivity will be altered and in other cases a Bootstrap effect will be evident. It is worth noting that monomers which are susceptible to strong medium effects will not have reliable Q-e values, a good example of this is 2-hydroxyethyl methacrylate (HEMA) where there is a large variation in reported values. The reactivity ratios of HEMA with STY have been reported to be strongly dependent on the medium,113 similarly the copolymerization of HEMA with lauryl methacrylate is solvent sensitive;114 behavior which has been attributed to non-ideal solution thermodynamics (cf Semchikov’s work in Section 13.2.3.4). Chapiro115 has published extensively on the formation of molecular associates in copolymerization involving polar monomers. Other common monomers which show strong solvent effects are N-vinyl-2-pyrrolidone, (meth)acrylic acids and vinyl pyridines. 13.2.3.4 Bootstrap model 13.2.3.4.1 Basic mechanism In the Bootstrap model, solvent effects on kp are attributed to solvent partitioning and the resulting difference between bulk and local monomer concentrations. In this way, a solvent could affect the measured kp without changing the reactivity of the propagation step. Bootstrap effects may arise from a number of different causes. As noted previously, when radical-solvent and monomer-solvent complexes form and the complexes do not propagate, the effect of complexation is to alter the effective radical or monomer concentrations, thereby causing a Bootstrap effect. Alternatively, a Bootstrap effect may arise from some bulk preferential sorption of one of the comonomers around the growing (and dead) polymer chains. This might be expected to occur if one of monomers is a poor solvent for its resulting polymer. A Bootstrap effect may also arise from a more localized from of preferential sorption, in which one of the monomers preferentially solvates the active chain end, rather than the entire polymer chain. In all cases, the result is the same: the effective free monomer and/or radical concentrations differ from those calculated from the monomer feed ratios, leading to a discrepancy between the predicted and actual propagation rates. 13.2.3.4.2 Copolymerization model Copolymerization models based upon a Bootstrap effect were first proposed by Harwood116 and Semchikov117 (see references cited therein). Harwood suggested that the terminal model could be extended by the incorporation of an additional equilibrium constant relating the effective and ‘bulk’ monomer feed ratios. Different versions of this so-called Bootstrap model may be derived depending upon the baseline model assumed (such as the terminal model or the implicit or explicit penultimate models) and the form of equilibrium expression used to represent the Bootstrap effect. In the simplest case, it is assumed that the magni-

792


tude of the Bootstrap effect is independent of the comonomer feed ratios. Hence in a bulk copolymerization, the monomer partitioning may be represented by the following equilibrium expression: f f1 = K 1bulk f2  f 2 bulk

   

The equilibrium constant K may be considered as a measure of the Bootstrap effect. Using this expression to eliminate the effective monomer fractions (f1 and f2) from the terminal model equations, replacing them with the measurable ‘bulk’ fractions (f1bulk and f2bulk), the following equations for composition43 and kp118 may be derived. F1 Kf1bulk r1Kf1bulk + f 2 bulk = F2 f 2 bulk r2 f 2 bulk + Kf1bulk kp =

r1K 2 f12bulk + 2Kf1bulk f 2 bulk + r2 f 22bulk 1 Kf1bulk + f 2 bulk r1Kf1bulk / k11 + r2 f 2 bulk / k 22

Examining the composition and kp equations above, it is seen that the Bootstrap effect K is always aliased with one of the monomer feed ratios (that is, both equations may be expressed in terms of Kf1 and f2). It is also seen that once Kf1 is taken as a single variable, the composition equation has the same functional form as the terminal model composition equation, but the kp equation does not. Hence it may seen that, for this version of the Bootstrap effect, the effect is an implicit effect - causing deviation from the terminal model kp equation only. It may also be noted that, if K is allowed to vary as a function of the monomer feed ratios, the composition equation also will deviate from terminal model behavior - and an explicit effect will result. Hence it may be seen that it is possible to formulate an implicit Bootstrap model (that mimics the implicit penultimate model) but in order to do this, it must be assumed that the Bootstrap effect K is constant as a function of monomer feed ratios. It should be noted that the above equations are applicable to a bulk copolymerization. When modelling solution copolymerization under the same conditions, the equations may be used for predicting copolymer composition since it is only the relationship between bulk and local monomer feed ratios that determines the effect on the composition and microstructure of the resulting polymer. However, some additional information about the net partitioning of monomer and solvent between the bulk and local phases is required before kp can be modelled. It should be observed that in a low-conversion bulk copolymerization, knowledge of the monomer feed ratios automatically implies knowledge of the individual monomer concentrations since, as there are no other components in the system, the sum of the monomer fractions is unity. However, in a solution copolymerization there is a third component - the solvent - and the monomer concentrations depend not only upon their feed ratio but also upon the solvent concentration. Modelling kp in a solution copolymerization could be achieved by re-writing the above equilibrium expression in terms of molar concentrations (rather than comonomer feed ratios), and including the solvent concentration in this expression.


793

The Bootstrap model may also be extended by assuming an alternative model (such as the explicit penultimate model) as the baseline model, and also by allowing the Bootstrap effect to vary as a function of monomer feed ratios. Closed expressions for composition and sequence distribution under some of these extended Bootstrap models may be found in papers by Klumperman and co-workers.43,44 13.2.3.4.3 Experimental evidence The Bootstrap model was introduced by Harwood,116 who studied three solvent sensitive copolymerizations (styrene/methacrylic acid, styrene/acrylic acid and styrene/acrylamide) and found that the copolymers of the same composition had the same sequence distribution irrespective of the solvent used. This meant that the conditional probabilities governing radical propagation were independent of the solvent. On this basis, he argued that composition and sequence distribution were deviating from their expected values because there was a difference between the monomer feed ratios in the vicinity of the active chain end, and those calculated on the basis of the bulk feed. In other words, the solvent was altering the rates of the individual propagation steps by affecting the reactant concentrations and not, as in the other solvent effects models, their reactivities. However, Fukuda et al.31 have argued that the NMR evidence provided by Harwood is not conclusive evidence for the Bootstrap model, since Harwood’s observations could also be described by variation of the reactivity ratios in such a way that their product (r1r2) remains constant. This has also been raised as an issue by Klumperman and O’Driscoll.43 They showed mathematically that a variation in the local comonomer ratio is not reflected in the monomer sequence distribution versus copolymer composition -this relationship being governed by the r1r2 product only. An alternative explanation for Harwood’s experimental data may be the stabilization or destabilization of the radicals by the solvent, an interpretation that would be analogous to the MCD model. Simple energy stabilization considerations, as used by Fukuda et al.119 to derive the penultimate unit effect, also suggest the constancy of r1r2. Prior to Harwood’s work, the existence of a Bootstrap effect in copolymerization was considered but rejected after the failure of efforts to correlate polymer-solvent interaction parameters with observed solvent effects. Kamachi,37 for instance, estimated the interaction between polymer and solvent by calculating the difference between their solubility parameters. He found that while there was some correlation between polymer-solvent interaction parameters and observed solvent effects for methyl methacrylate, for vinyl acetate there was none. However, it should be noted that evidence for radical-solvent complexes in vinyl acetate systems is fairly strong (see Section 3), so a rejection of a generalized Bootstrap model on the basis of evidence from vinyl acetate polymerization is perhaps unwise. Kratochvil et al.120 investigated the possible influence of preferential solvation in copolymerizations and concluded that, for systems with weak non-specific interactions, such as STY-MMA, the effect of preferential solvation on kinetics was probably comparable to the experimental error in determining the rate of polymerization (±5%). Later, Maxwell et al.121 also concluded that the origin of the Bootstrap effect was not likely to be bulk monomer-polymer thermodynamics since, for a variety of monomers, Flory-Huggins theory predicts that the monomer ratios in the monomer-polymer phase would be equal to that in the bulk phase.122 Nevertheless, there are many copolymerization systems for which there is strong evidence for preferential solvation, in particular, polymer solutions exhibiting cosolvency or where one of the solvents is a non-solvent for the polymer. Preferential adsorption and

794


desorption are manifest where the polymer adjusts its environment towards maximum solvation. With this knowledge it may be expected that Bootstrap effects based on preferential solvation will be strongest where one of the monomers (or solvents) is a poor or non-solvent for the polymer (such as copolymerization of acrylonitrile or N-vinyl carbazole). Indeed, early experimental evidence for a partitioning mechanism in copolymerization was provided by Ledwith et al.123 for the copolymerization of N-vinyl carbazole with MMA in the presence of a range of solvents. Direct evidence for preferential solvation was obtained by Semchikov et al.,124 who suggested that it could be detected by calculating, from measurements of the solution thermodynamics, the total and excess thermodynamic functions of mixing. Six monomer pairs were selected -Vac-NVP, AN-STY, STY-MA, Vac-STY, STY-BMA and MMA-STY. The first four of these monomer pairs were known to deviate from the terminal model composition equation, while the latter two were not. They found that these first four copolymerizations had positive ∆GE values over the temperature range measured, and thus also formed non-ideal polymer solutions (that is, they deviated from Raoult’s law). Furthermore, the extent of deviation from the terminal model composition equation could be correlated with the size of the ∆GE value, as calculated from the area between the two most different composition curves obtained for the same monomer pair under differing reaction conditions (for example, initiator concentration; or type and concentration of transfer agent). For STY-MMA they obtained negative ∆GE values over the temperature range considered, but for STY-BMA negative values were obtained only at 318 and 343K, and not 298K. They argued that the negative ∆GE values for STY-MMA confirmed the absence of preferential solvation in this system, and hence its adherence to the terminal model composition equation. For STY-BMA they suggested that non-classical behavior might be expected at low temperatures. This they confirmed by polymerizing STY-BMA at 303K and demonstrating a change in reactivity ratios of STY-BMA with the addition of a transfer agent. Based upon the above studies, it may be concluded that there is strong evidence to suggest that Bootstrap effects arising from preferential solvation of the polymer chain operate in many copolymerization systems, although the effect is by no means general and is not likely to be significant in systems such as STY-MMA. However, this does not necessarily discount a Bootstrap effect in such systems. As noted above, a Bootstrap effect may arise from a number of different phenomena, of which preferential solvation is but one example. Other causes of a Bootstrap effect include preferential solvation of the chain end, rather than the entire polymer chain,121,125 or the formation of non-reactive radical-solvent or monomer-solvent complexes. In fact, the Bootstrap model has been successfully adopted in systems, such as solution copolymerization of STY-MMA, for which bulk preferential solvation of the polymer chain is unlikely. For instance, both Davis125 and Klumperman and O’Driscoll43 adopted the terminal Bootstrap model in a reanalysis of the microstructure data of San Roman et al.126 for the effects of benzene, chlorobenzene and benzonitrile on the copolymerization of MMA-STY. Versions of the Bootstrap model have also been fitted to systems in which monomer-monomer complexes are known to be present, demonstrating that the Bootstrap model may provide an alternative to the MCP and MCD models in these systems. For instance, Klumperman and co-workers have successfully fitted versions of the penultimate Bootstrap model to the systems styrene with maleic anhydride in butanone and toluene,43 and styrene


795

with acrylonitrile in various solvents.44 This latter work confirmed the earlier observations of Hill et al.127 for the behavior of styrene with acrylonitrile in bulk, acetonitrile and toluene. They had concluded that, based on sequence distribution data, penultimate unit effects were operating but, in addition, a Bootstrap effect was evident in the coexistent curves obtained when triad distribution was plotted against copolymer composition for each system. In the copolymerization of styrene with acrylonitrile Klumperman et al.44 a variable Bootstrap effect was required to model the data. Given the strong polarity effects expected in this system (see Section 13.2.2), part of this variation may in fact be caused by the variation of the solvent polarity and its affect on the reactivity ratios. In any case, as this work indicates, it may be necessary to simultaneously consider a number of different influences (such as, for instance, penultimate unit effects, Bootstrap effects, and polarity effects) in order to model some copolymerization systems. 13.2.4 CONCLUDING REMARKS Solvents affect free-radical polymerization reactions in a number of different ways. Solvent can influence any of the elementary steps in the chain reaction process either chemically or physically. Some of these solvent effects are substantial, for instance, the influence of solvents on the gel effect and on the polymerization of acidic or basic monomers. In the specific case of copolymerization then solvents can influence transfer and propagation reactions via a number of different mechanisms. For some systems, such as styrene-acrylonitrile or styrene-maleic anhydride, the selection of an appropriate copolymerization model is still a matter of contention and it is likely that complicated copolymerization models, incorporating a number of different phenomena, are required to explain all experimental data. In any case, it does not appear that a single solvent effects model is capable of explaining the effect of solvents in all copolymerization systems, and model discrimination should thus be performed on a case-by-case basis. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

Bednarek, D., Moad, G., Rizzardo, E. and Solomon, D. H., Macromolecules, 1988, 21, 1522. Kiefer, H. and Traylor, T., J. Am. Chem. Soc., 1967, 89, 6667. Terazima, M., Tenma, S., Watanabe, H. and Tominaga, T., J. Chem. Soc.-Farad. Trans., 1996, 92, 3057. Terazima, M. and Hamaguchi, H., J. Phys. Chem., 1995, 99, 7891. Bamford, C. H. and Brumby, S., Makromol. Chem., 1967, 105, 122. Burnett, G. M., Cameron, G. G. and Joiner, S. N., J. Chem. Soc., Faraday Trans., 1972, 69, 322. Burnett, G. M., Dailey, W. S. and Pearson, J. M., Trans. Faraday Soc., 1972, 61, 1216. Anderson, D. B., Burnett, G. M. and Gowan, A. C., J. Polym. Sci. Part A: Polym. Chem., 1963, 1, 1456. Zammit, M. D., Davis, T. P., Willett, G. D. and O’Driscoll, K. F., J. Polym. Sci: Part A: Polym. Chem., 1997, 35, 2311. Shtamm, E. V., Skurlatov, Y. I., Karaputadse, I. M., Kirsh, Y. E. and Purmal, A. P., Vysokomol. Soedin.B, 1980, 22, 420. Pascal, P., Napper, D. H., Gilbert, R. G., Piton, M. C. and Winnik, M. A., Macromolecules, 1990, 23, 5161. Morrison, D. A. and Davis, T. P., Macromol. Chem. Phys., 2000 - submitted. Davis, T. P., O’Driscoll, K. F., Piton, M. C. and Winnik, M. A., Macromolecules, 1989, 22, 2785. Coote, M. L. and Davis, T. P., Eur. Polym. J., 2000-in press. Olaj, O. F. and Schnoll-Bitai, I., Monatsh. Chem., 1999, 130, 731. Beuermann, S., Buback, M., Schmaltz, C. and Kuchta, F.-D., Macromol. Chem. Phys., 1998, 199, 1209. Yamada, K., Nakano, T. and Okamoto, Y., Macromolecules, 1998, 31, 7598. Kamachi, M., Liaw, D. J. and Nozakura, S., Polym. J., 1979, 12, 921. Coote, M. L. and Davis, T. P., Progr. Polym. Sci., 2000, 24, 1217. Renaud, P. and Gerster, M., Angew. Chem. Int. Ed., 1998, 37, 2562. Odian, G., Principles of Polymerization; 2nd edn. Wiley, New York, 1981. Harrisson, S., Kapfenstein, H. M. and Davis, T. P., Macromolecules, 2000- submitted.

796

23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74

Michelle L. Coote and Thomas P. Davis Heuts, J. P. A., Forster, D. J. and Davis, T. P., Macromolecules, 1999, 32, 3907. Davis, T. P., Haddleton, D. M. and Richards, S. N., J. Macromol. Sci., Rev. Macromol. Chem. Phys., 1994, C34, 234. Benson, S. W. and North, A. M., J. Am. Chem. Soc., 1962, 84, 935. Mahabadi, H. K. and O’Driscoll, K. F., J. Polym. Sci., Polym. Chem. Ed., 1977, 15, 283. Norrish, R. G. W. and Smith, R. R., Nature (London), 1942, 150, 336. O’Neil, G. A., Wisnudel, M. B. and Torkelson, J. M., Macromolecules, 1998, 31, 4537. Cameron, G. G. and Cameron, J., Polymer, 1973, 14, 107. Mayo, F. R. and Lewis, F. M., J. Am. Chem. Soc., 1944, 66, 1594. Fukuda, T., Kubo, K. and Ma, Y.-D., Prog. Polym. Sci., 1992, 17, 875. Fukuda, T., Ma, Y.-D. and Inagaki, H., Macromolecules, 1985, 18, 17. see for example(a) Wong, M. W., Pross, A., Radom, L., J. Am. Chem. Soc., 1994, 116, 6284; (b) Wong, M., W., Pross, A.; Radom, L., J. Am. Chem. Soc., 1994, 116, 11938. For an account of solvent effects in chemical reactions see for example: Pross, A. Theoretical and Physical Principles of Organic Reactivity; John Wiley & Sons, Inc.: New York, 1995. Price, C. C., J. Polym. Sci., 1946, 1, 83. Mayo, F. R. and Walling, C., Chem. Rev., 1950, 46, 191. Kamachi, M., Adv. Polym. Sci., 1981, 38, 56. Lewis, F., Walling, C., Cummings, W., Briggs, E. R. and Mayo, F. R., J. Am. Chem. Soc., 1948, 70, 1519. Nozaki, K., J. Polym. Sci., 1946, 1, 455. Sandner, B. and Loth, E., Faserforsch Textiltechnology, 1976, 27, 571. Sandner, B. and Loth, E., Faserforsch. Textiltechnology, 1976, 27, 633. See for example: (a) Zytowski, T., Fischer, H., J. Am. Chem. Soc., 1996, 118, 437; (b) Heberger, K., Fischer, H., Int. J. Chem. Kinetics, 1993, 25, 249. Klumperman, B. and O’Driscoll, K. F., Polymer, 1993, 34, 1032. Klumperman, B. and Kraeger, I. R., Macromolecules, 1994, 27, 1529. Russell, G. A., J. Am. Chem. Soc., 1958, 80, 4897. Russell, G. A., Tetrahedron, 1960, 8, 101. Russell, G. A., Ito, K. and Hendry, D. G., J. Am. Chem. Soc., 1961, 83, 2843. Sadhir, R. K., Smith, J. D. B. and Castle, P. M., J. Polym. Sci: Part A: Polym. Chem., 1985, 23, 411. Strong, R. L., Rand, S. J. and Britl, A. J., J. Am. Chem. Soc., 1960, 82, 5053. Strong, R. L. and Perano, J., J. Am. Chem. Soc., 1961, 83, 2843. Strong, R. L., J. Phys. Chem., 1962, 66, 2423. Martin, M. M. and Gleicher, G. J., J. Am. Chem. Soc., 1964, 86, 238. Russell, G. A., J. Org. Chem., 1959, 24, 300. Burnett, G. M., Cameron, G. G. and Cameron, J., Trans. Faraday Soc., 1973, 69, 864. Buchachenko, A. L., Sukhanova, O. P., Kalashnikova, L. A. and Neiman, M. B., Kinetika i Kataliz, 1965, 6, 601. Kalashnikova, L. A., Neiman, M. B. and Buchachenko, A. L., Zh. Fiz. Khim., 1968, 42, 598. Kalashnikova, L. A., Buchachenko, A. L., Neiman, M. B. and Romantsev, E. G., Zh. Fiz. Khim., 1969, 43, 31. Hatada, K., Kitayama, T. and Yuki, H., Makromol. Chem., Rapid Commun., 1980, 1, 51. Busfield, W. K., Jenkins, I. D. and Monteiro, M. J., Aust. J. Chem., 1997, 50, 1. Busfield, W. K., Jenkins, I. D. and Monteiro, M. J., J. Polym. Sci.,;Part A, Polym. Sci., 1997, 35, 263. Busfield, W. K., Jenkins, I. D. and Monteiro, M. J., Polymer, 1996, 38, 165. Ratzsch, M. and Vogl, O., Progr. Polym. Sci., 1991, 16, 279. Mayo, F., J. Am. Chem. Soc., 1958, 80, 4987. Burnett, G. M., Dailey, W. S. and Pearson, J. M., Trans. Faraday Soc., 1965, 61, 1216. Burnett, G. M., Dailey, W. S. and Pearson, J. M., Eur. Polym. J., 1969, 5, 231. Hall Jr., H. K. and Daly, R. C., Macromolecules, 1975, 8, 23. Henrici-Olive, G. and Olive, S., Makromol. Chem., 1963, 68, 219. Henrici-Olive, G. and Olive, S., Z. Phys. Chem., 1965, 47, 286. Henrici-Olive, G. and Olive, S., Z. Phys. Chem., 1966, 48, 35. Henrici-Olive, G. and Olive, S., Z. Phys. Chem., 1966, 48, 51. Henrici-Olive, G. and Olive, S., Makromol. Chem., 1966, 96, 221. Burnett, G. M., Cameron, G. G. and Zafar, M. M., Eur. Polym. J., 1970, 6, 823. Burnett, G. M., Cameron, G. G. and Joiner, S. N., J. Chem. Soc., Faraday Trans., 1973, 69, 322. Kamachi, M., Satoh, J. and Nozakura, S.-I., J. Polym. Sci. Polym. Chem. Ed., 1978, 16, 1789.


75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128

797

Pascal, P., Napper, D. H., Gilbert, R. G., Piton, M. C. and Winnik, M. A., Macromolecules, 1993, 26, 4572. Heublein, B. and Heublein, G., Acta Polym., 1988, 39, 324. O’Driscoll, K. F. and Monteiro, M. J., 1996. Czerwinski, W. K., Macromolecules, 1995, 28, 5411. Bartlett, P. D. and Nozaki, K., J. Am. Chem. Soc., 1946, 68, 1495. Seiner, J. A. and Litt, M., Macromolecules, 1971, 4, 308. Cais, R. E., Farmer, R. G., Hill, D. J. T. and O’Donnell, J. H., Macromolecules, 1979, 12, 835. Bamford, C. H. and Barb, W. G., Discuss. Faraday Soc., 1953, 208. Barb, W. G., J. Polym. Sci., 1953, 11, 117. Tsuchida, E. and Tomono, H., Makromol. Chem., 1971, 141, 265. Tsuchida, E., Tomono, T. and Sano, H., Makromol. Chem., 1972, 151, 245. Dodgson, K. and Ebdon, J. R., Makromol. Chem., 1979, 180, 1251. Dodgson, K. and Ebdon, J. R., Eur. Polym. J., 1977, 13, 791. Walling, C., Briggs, E. R., Wolfstern, K. B. and Mayo, F. R., J. Am. Chem. Soc., 1948, 70, 1537. Farmer, R. G., Hill, D. J. T. and O’Donnell, J. H., J. Macromol. Sci., Chem., 1980, A14, 51. Hill, D. J. T., O’Donnell, J. J. and O’Sullivan, P. W., Prog. Polym. Sci., 1982, 8, 215. Sanayei, R. A., O’Driscoll, K. F. and Klumperman, B., Macromolecules, 1994, 27, 5577. Litt, M., Macromolecules, 1971, 4, 312. Barb, W. G., Proc. Roy. Soc., Ser. A, 1952, 212, 66. Barb, W. G., J. Polym. Sci., 1952, 10, 49. Booth, D., Dainton, F. S. and Ivin, K. J., Trans. Faraday Soc., 1959, 55, 1293. Lewis, F. M., Walling, C., Cummings, W., Briggs, E. R. and Wenisch, W. J., J. Am. Chem. Soc., 1948, 70, 1527. Hill, D. J. T., O’Donnell, J. H. and O’Sullivan, P. W., Macromolecules, 1982, 15, 960. Cowie, J. M. G., in Alternating Copolymers; Vol. , ed. Cowie, J. M. G., Plenum, New York, 1985, . Seymour, R. B. and Garner, D. P., Polymer, 1976, 17, 21. Seymour, R. B. and Garner, D. P., Polym. News, 1978, 4, 209. Klumperman, B. “Free Radical Copolymerization of Styrene and Maleic Anhydride,” PhD Thesis, Technische Universiteit Eindhoven, 1994. Arnaud, R., Caze, C. and Fossey, J., J. Macromol. Sci. - Chem., 1980, A14, 1269. Iwatsuki, S. and Yamashita, Y., Makromol. Chem., 1967, 104, 263. Olson, K. G. and Butler, G. B., Macromolecules, 1983, 16, 710. Rätzsch, M. and Seinert, V., Makromol. Chem., 1984, 185, 2411. Butler, G. B., Olson, K. G. and Tu, C.-L., Macromolecules, 1984, 17, 1884. Smirnov, A. I., Deryabina, G. L., Kalabina, A. L., Petrova, T. L., Stoyachenko, I. L., Golubev, V. B. and Zubov, V. P., Polym. Sci. USSR (Engl. Transl.), 1978, 20, 2014. Golubev, V. B., Zubov, V. P., Georgiev, G. S., Stoyachenko, I. L. and Kabanov, V. A., J. Polym. Sci., Polym. Chem. Ed., 1973, 11, 2463. Barton, J., Capek, I. and Tino, J., Makromol. Chem., 1980, 181, 255. Karad, P. and Schneider, C., J. Polym. Sci. Part A: Polym. Chem., 1983, 16, 1295. Hill, D. J. T., O’Donnell, J. H. and O’Sullivan, P. W., Macromolecules, 1983, 16, 1295. Giese, B. and Meister, J., Chem. Ber., 1977, 110, 2558. Lebduska, J., Snuparek, J., Kaspar, K. and Cermak, V., J. Polym. Sci: Part A: Polym. Chem., 1986, 24, 777. Ito, K., Uchida, K., Kitano, T., Yamada, E. and Matsumoto, T., Polym. J., 1985, 17, 761. Chapiro, A. and Perec-Spitzer, L., Eur. Polym. J., 1975, 25, 713. Harwood, H. J., Makromol. Chem. Makromol. Symp., 1987, 10/11, 331. Semchikov, Y. D., Macromol. Symp., 1996, 111, 317. Coote, M. L., Johnston, L. P. M. and Davis, T. P., Macromolecules, 30, 8191 (1997). Fukuda, T., Ma, Y.-D. and Inagaki, H., Makromol. Chem., Rapid Commun., 1987, 8, 495. Kratochvil, P., Strakova, D., Stejskal, J. and Tuzar, Z., Macromolecules, 1983, 16, 1136. Maxwell, I. A., Aerdts, A. M. and German, A. L., Macromolecules, 1993, 26, 1956. Maxwell, I. A., Kurja, J., Doremaele, G. H. J. v. and German, A. L., Makromol Chem., 1992, 193, 2065. Ledwith, A., Galli, G., Chiellini, E. and Solaro, R., Polym. Bull., 1979, 1, 491. Egorochkin, G. A., Semchikov, Y., D., Smirnova, L. A., Karayakin, N. V. and Kut’in, A. M., Eur. Polym. J., 1992, 28, 681. Davis, T. P., Polym. Commun., 1990, 31, 442. San Roman, J., Madruga, E. L. and Puerto, M. A., Angew Makromol. Chem., 1980, 86, 1. Hill, D. J. T., Lang, A. P., Munro, P. D. and O’Donnell, J. H., Eur. Polym. J., 1992, 28, 391. Coote, M. L., Davis, T. P., Klumperman, B. and Monteiro, M. J., J.M.S.-Rev. Macromol. Chem. Phys., 1998, C38, 567.

798

Maw-Ling Wang

13.3 EFFECTS OF ORGANIC SOLVENTS ON PHASE-TRANSFER CATALYSIS Maw-Ling Wang Department of Chemical Engineering National Chung Cheng University, Taiwan, ROC

The reaction of two immiscible reactants is slow due to their low solubilities and limited contact surface area. The conventional way to improve the reaction rate or to elevate the conversion of reactants is to increase the agitation speed, temperature, or use the protic or aprotic solvent to dissolve the reactants. The increase in agitation speed can increase the contact surface area between two phases only to a certain value. Thus, the reaction rate or the conversion is limited by the increase in the agitation speed. Usually, the rate of reaction is increased by raising the temperature. However, byproducts are accompanied by elevating the solution temperature. The separation of product from byproducts or catalyst makes the cost to increase. Although protic solvent (CH3OH, or CH3COOH) can dissolve reactants, solvation and hydrogen bonding make the activity of the nucleophilic anion decrease significantly. Thus, the reaction rate using protic solvent is retarded. For the other case, the reaction rate is largely increased using aprotic solvent. The application of aprotic solvent is also limited because of cost and recovery difficulty. For this, the problem of two-phase reaction is not overcome until the development of phase-transfer catalysis (PTC). Phase-transfer catalytic reactions provide an effective method in organic synthesis from two immiscible reactants in recent development.93,103,111,113 In 1951, Jarrouse47 found that the reaction of aqueous-soluble sodium cyanide (NaCN) and organic-soluble 1-chlorooctane (1-C8H17Cl) is dramatically enhanced by adding a small amount of quaternary ammonium salt (R4N+X-, or Q+X-, Q+: R4N+). The reaction is almost complete and a 95% conversion is obtained within two hours when a catalytic amount of tetra-n-butylammonium chloride ((C4H9)4N+Cl-, or Q+Cl-, Q+: (C4H9)4N+) is added. The mechanism of the reaction of sodium cyanide and 1-chlorooctane in organic solvent/water two-phase medium is expressed as

[13.3.1]

As shown in Equation [13.3.1], sodium cyanide (NaCN) and 1-chlorooctane (1-C8H17Cl) are soluble in aqueous phase and organic phase, respectively. In the aqueous phase, NaCN first reacts with tetra-n-butylammonium chloride ((C4H9)4N+Cl-, Q+Cl-) to produce organic-soluble tetra-n-butylammonium cyanide ((C4H9)4N+CN-, Q+CN-). Then, this tetra-n-butylammonium cyanide (QCN) further reacts with 1-chlorooctane (1-C8H17Cl) to produce 1-cyanooctane (C8H17CN) in the organic phase. Tetra-n-butylammonium chloride ((C4H9)4N+Cl-), which is also produced from the organic-phase reaction, transfers to the aqueous phase, prepared for further regeneration. It is obvious that PTC reaction107 involves

13.3 Effects of organic solvents on phase-transfer catalysis

799

a two-phase reaction (aqueous-phase and organic-phase reaction), transfer of QCN from aqueous phase to organic phase and transfer of QCl from organic phase to aqueous phase, and equilibrium partition of QCN and QCl between organic and aqueous phases, respectively. The overall reaction rate highly depends on the intrinsic rate constants in aqueous phase and organic phase, the mass transfer rate of QCN and QBr, and the equilibrium partition coefficient of QCN and QBr, which are all affected by the interaction of components and their environments. The organic solvent provides the environment for the interaction of reactants. Therefore, the organic solvent plays an important role in influencing the reaction rate and the conversion of reactant. Since then, Makosza used an interfacial mechanism65-67 to describe the behavior in the two-phase reaction. Later, Starks107 used the extraction mechanism to explain the behavior in the two-phase reaction and selected phase-transfer catalysis (PTC) to describe this special chemical process.14,161 The most important advantage of using PTC technique is in synthesizing specialty chemicals with almost no byproducts and moderate reaction conditions. Today, PTC is widely applied to various reactions via substitution, displacement, condensation, oxidation and reduction, polymer modification and polymerization to synthesize specialty chemicals. Based on the reaction mechanism, phase-transfer catalysis can be classified as: (1) normal phase-transfer catalysis (NPTC), (2) reverse phase-transfer catalysis (RPTC), and (3) inverse phase-transfer catalysis (IPTC). Equation [13.3.1] illustrates the typical reaction for NPTC. The phase-transfer catalyst (Q+) brings the nucleophilic reagent (CN-) from aqueous phase to organic phase. Quaternary ammonium salts, quaternary phosphonium salts, crown ethers, polyethylene glycols (PEGs) and tertiary amines are the common normal phase-transfer catalyst (NPTC).17,29,94,108,109,110,128,130,152 In general, the cation transfers from aqueous phase to organic phase in the RPTC. The principle of reverse phase-transfer catalysis (RPTC)24,42-44,50 is that an ion pair is formed from catalyst and cation in the aqueous phase. This ion-paired compound then transfers to the organic phase reacting with an organic-phase reactant. Alkyl-aryl sulfonate (RSO3Na), such as sodium 4-dodecylbenzene sulfonate (NaDBS) and tetraarylboronate such as sodium tetra(diperfluoromethyl)phenyl-boronate (TFPB) are the common reverse phase-transfer catalysts. However, few results were reported using reverse phase-transfer catalysis in synthesizing specialty chemicals.24,42-44,50 A typical reaction mechanism in a liquid-liquid two-phase solution is given by Equation [13.3.2] In the NPTC and RPTC, the function of the catalyst is that it first reacts with aqueous-phase reactant to produce an organic-soluble ion-pair compound. Mathias and Vaidya69 found that an aqueous-soluble ion pair was produced in the organic phase from the reaction of alanine and benzoyl chloride catalyzed by 4-dimethylaminopyridine (DMAP). This discovery initiated the research of the field in inverse phase-transfer catalysis (IPTC), in which the catalyst first reacts with organic-phase reactant in the organic phase to produce an aqueous-soluble ion-paired intermediate. Then, this aqueous-soluble ion-paired intermediate transfers to the aqueous phase, prepared for reacting with aqueous-phase reactant to produce the desired product. Catalyst is released in the aqueous phase and transferred to the organic phase for further regeneration. A typical IPTC mechanism of the reaction of benzoyl chloride and sodium acetate to synthesize ester compound in the liquid-liquid two-phase reaction is expressed by Equation [13.3.3]

800

Maw-Ling Wang

[13.3.2]

Inverse phase-transfer catalysis (IPTC) can be applied to synthesize symmetric and antisymmetric acid anhydride in organic synthesis.26,54,69,102,148-150,153 Pyridine 1-oxide (PNO), 4-dimethylaminopyridine (DMAP), 4-pyrrolidinopyridine (PPY) and 1-methyl-2(1H)-pyridothione are usually used as the inverse phase-transfer catalysts (IPTC).159

[13.3.3]

The characteristics of two-phase phase-transfer catalytic (PTC) systems are: the presence of at least two phases and at least one interfacial region separating the phases.24,26,42,43,50,69,107,109,110,152 The reactions involve: (1) transfer of an ion or compound from its normal phase into the reaction phase or interfacial region, (2) reaction of the transferred ion or compound with the non transferred reactant located in the reaction phase or interfacial region, and (3) transfer of the product from the reaction phase or interfacial region into its normal phase. For example, a successful NPTC process involves (1) the maximization of the rate of transfer of reactant anions from the aqueous or solid phase to the organic phase, (2) the maximization of the rate of transfer of product anions from the organic phase to the aqueous phase or solid phase, and (3) the related equilibrium partitioning of the reactant and product anions between organic and aqueous or solid phases. The anion must not only transfer to the organic phase, but once there the anion must be in a highly reactive form. Some organic-phase reactions are so fast that the transferred anion requires little or no activation beyond just being delivered to the organic phase. Other reactions require substantial anion activation before useful and practical reaction rates can be achieved. It is obvious that the polarity of the organic solvent affects the activation of the anion as well as the difference


801

in the cation-anion interionic distance for the two ion pairs. In principle, anions do not have a great affinity for nonpolar solvent and prefer to reside in an aqueous phase. Ease or difficulty of transfer of most anions into organic-phase solution is also highly affected by the organic solvent, i.e., interaction of the organic solvent and the reactant. In general, a polar solvent may be necessary to obtain an appropriate rate of the anion transfer to the organic phase for a NPTC process. Solvent may be necessary to increase the rate of the organic-phase reaction. The most common solvent, dichloromethane (CH2Cl2), has been extensively used as a polar solvent in the PTC work because it readily dissolves most quaternary salts and other phase-transfer agents, and because it is polar to speed both the transfer step and the organic-phase reaction step. Although the hydrocarbons suffer from lack of polarity, they have also been extensively used as solvents for PTC systems. The main reasons are that they are reasonably safe, inexpensive and easy to recover in a high purity. One strategy for selecting organic solvent is that a high boiling point solvent is selected for a reaction in which the product has a low-boiling point. In other cases, a solvent might be chosen to minimize solubility of phase-transfer agent in the organic phase to force formation of third phase (catalyst) from which the phase-transfer catalyst may be more easily separated or extracted. The phase-transfer catalytic reactions (NPTC, RPTC or IPTC) are usually carried out in a liquid-liquid two-phase medium. They have been extensively applied to liquid-gas, liquid-solid two-phase media.18,21,63,128-130 However, purification of product from catalyst in the liquid phase of a final solution is difficult to produce a product of high purity. In 1975, Regen and coworkers88-92 proposed triphase catalysis (TC) in which the catalyst is immobilized on a porous solid support (usually polymer). The solid catalyst is easily separated from the final products after reaction by mechanical separation processes, such as centrifugation or filtration. The organophilicity and the hydrophilicity of the solid polymer support greatly influence the content and the imbibed composition of the organic phase and the aqueous phase within the solid porous polymer support. Hence, the reaction rates are determined by the concentrations of reactants in both the organic phase and aqueous phase, they are controlled by the organic solvents. Therefore, it is important to understand the characteristics of the organic phase in the triphase catalyst as well as the characteristics of the organic solvent in the liquid-liquid two-phase PTC reaction. 13.3.1 TWO-PHASE PHASE-TRANSFER CATALYTIC REACTIONS 13.3.1.1 Theoretical analysis of the polarity of the organic solvents and the reactions The transfer of anions from an aqueous phase to an organic phase may be achieved by choosing a phase-transfer cation that is not strongly solvated by water, that has organic-like characteristics, and is compatible with the organic phase for NPTC. The factors that affect the mass transfer and the distribution of the phase-transfer catalyst cation-anion pair between the organic and aqueous phases include: (1) the charge-to-volume ratio of the anion, the polarizability, and the organic structure of the cation associated with the anion, (2) the hydrophilic-organophilic balance of the associated cation; (3) the polarity of the organic phase; (4) the hydration of the anion; (5) the presence of aqueous salts and/or aqueous hydroxide ions.

802

Maw-Ling Wang

Both cation and anion of phase-transfer catalyst can affect the distribution of the PTC between two phases, and hence the reaction rate. The partitioning equilibrium of the anion between organic and aqueous phases can be qualitatively estimated from the free energies of the anion transfer from water to organic phase. A large positive free energy of transfer from the aqueous phase to the organic phase clearly indicates that the anion prefers to reside in the aqueous phase. For example, the free energies of transfer of Cl-, Br-, and I- from water to acetonitrile are +11.6, +8.1 and +4.8 Kcal/mol, respectively.110 Thus, the transfer from aqueous to organic phase becomes less unfavorable as one proceeds from chloride to bromide to iodide. These trends may be understood in terms of the change in charge-to-volume ratios of the halide ions. Because chloride has the largest charge-to-volume ratio, it is the least polarizable and the most strongly hydrated. In contrast, iodide has a relatively diffuse charge and is less strongly hydrated. A successful phase-transfer catalytic reaction occurs when the process is able to transfer the anions from the aqueous phase to the organic phase or vice versa for the reaction to proceed, and the transferred anions are active and prepared for reaction. An active catalyst needs to be sufficiently distributed in the organic phase for the reaction to occur. The distribution of catalysts and the associated anion in the organic phase strongly depends on the structure of the quaternary cation and the hydration of anion being transferred into the aqueous phase. Therefore, the following results are used for the reference in selective NPTC catalyst. (1) Tetramethylammonium cation with a simple anion (CH3)4N+Y- (Y=Cl-, Br-, CN-, etc.) is not easily distributed in most organic solutions. Therefore, (CH3)4N+Y- are usually not good PTCs. The only ways to increase the distribution of (CH3)4N+Y- is to couple the cation with a large organic anion13,49,78,95,155,156 or to use an organic solvent of high purity.17,107,109,110,152 (2) Tetraethylammonium ((C2H5)4N+Y-) and tetrapropylammonium ((C3H7)4N+Y-) salts are also poor catalysts for transferring small anions into most organic solutions.51,105 (3) Tetrabutylammonium salts show high efficiencies as phase-transfer catalysts. They are readily available in high purity on a commercial scale. (4) Quaternary ammonium cations, R4N+, R=C5H11 to C10H21 easily extract anions into organic phase and exhibit higher catalytic activities. (5) Higher tetraalkylammonium salts, R4N+X, R: (C12H25)4N+ and higher groups, can easily extract anion into an organic phase. However, the interchange of anions between organic and aqueous phases is slow and the reaction rate decreases compared with quaternary salts where R = C5H11 to C10H21. Table 13.3.1 shows the effect of catalyst structure on the rate of PTC reaction of thiophenoxide with 1-bromooctane. In addition to the preference of anion to reside in the aqueous or organic phase, a distribution ratio (or partition coefficient), α, of phase-transfer catalyst (QX) cation between aqueous and organic phase is defined as α = [QX]org/[QX]aq

[13.3.4]

Use of solvents having higher polarity facilitates distribution of quaternary salts into organic solvents. Hence, it also allows use of smaller quaternary salts as catalysts. With di-


803

Table 13.3.1. Effect of catalyst chloromethane or solvent of a similar polarity, it structure on the rate of the reaction is possible to use tetramethylammonium cation, of thiophenoxide and 1-bromooctane (CH ) N+, or tetrapropylammonium cation, 3 4 in benzene/water solution + k×103 , Lmol-1s-1

Catalyst (CH3)4N+Br+

-

+

-

10.0 C6H6 all catalyst cation-anion pairs to be parti>10.0 C6H5Cl tioned into the organic phase. This behavior >10.0 o-C6H4Cl2 is evidenced in the cyanide displacement on 1-bromooctane catalyzed by tetra-n>10.0 n-C4H9Cl butylphosphonium bromide ((C4H9)4P+Br-, 3.33 Cl(CH2)4Cl or Q+Br-).107,109,110,111,114 The catalyst is only sparingly soluble in 1-bromooctane, but is 0.34 Cl(CH2)3Cl substantially more soluble in aqueous so0.16 Cl(CH2)2Cl dium cyanide solution, so that initially little 0.028 ClCH2Cl Q+CN- is in the organic phase and the displacement reaction is slow. However, 0.021 CHCl3 tetra-n-butylphosphonium salts ((C4H9)4 0.91 C2H5COC2H5 P+CN-, Q+CN-) are more soluble in the 0.071 CH3COC2H5 product 1-cyanooctane; therefore, as the conversion of 1-bromooctane to 0.014 n-C4H9OH 1-cyanooctane continues, increasing quantities of the catalyst are taken into the or*EQBr= [Q+]aq[Br-]aq/[QBr]org ganic phase, and the reaction rate Data adopted from Brandstrom4 accelerates. This behavior signals the autocatalytic character of reaction. Not only does the solvent affect the reaction rate, but it also determines the reaction mechanism. In Starks’ extraction mechanism of PTC, most reacting compound transfers to the bulk phase. However, reaction may occur at the interface of the two phases. For example: hexachlorocyclotriphosphazene has been reported to react very slowly with 2,2,2-trifluoroethanol in an alkaline solution of NaOH/C6H5Cl two-phase system in the absence of phase-transfer catalyst.136-140 Since sodium 2,2,2-trifluoroethanoxide is not soluble in chlorobenzene, the process probably proceeds at the interface region of the system. Similar is the reaction of benzylation of isobutyraldehyde in the presence of tetra-n-butylammonium iodide in an alkaline solution of NaOH/toluene, which is a two-phase system.37 Makosza interfacial mechanism65-67 was employed to rationalize the experimental results. The main reason is that the ammonium salt of the nucleophilic reagent is not soluble in toluene. Usually, the nucleophilic substitutions under NPTC condition are described by an SN2-type reaction both in solid-liquid and liquid-liquid systems in which they can proceed at the interface through the formation of cyclic adsorption complexes.160 The activity of the nucleophilic reagent in the organic phase is determined by the polarity of the organic solution and the hydration in liquid-liquid system. In the solid-liquid system, the reaction is highly affected by the organic solvent. Solvent

Extraction constant, EQBr*


805

13.3.1.2 Effect of organic solvent on the reaction in various reaction systems (A) Synthesis of ether compound catalyzed by quaternary ammonium salts (NPTC) One of the most useful synthesis applications of phase transfer catalysis (PTC) is in the preparation of ether according to the following general equation R′X + ROX + OH −  PTC  → ROR′ + H2 O + X −

[13.3.6]

where R and R' are the primary or secondary alkyl or aryl groups, X is a halide and the caustic base is usually sodium or potassium hydroxide in the aqueous solution. The generally accepted reaction mechanism is

[13.3.7]

It is important to consider that the alkoxide ion (RO-) is a reactive nucleophilic but also a strong base. It was shown that 10-3 M C6H5(CH3)3N+OC4H9 is 1000 times more basic than KOC4H9 (both in C4H9OH). Extracted alkoxide bases can be applied in principle to numerous base-catalyzed reactions, e.g., oxidations, eliminations and isomerization. Better quantitative understanding of the extraction of alkoxide into organic phase is important. Dehmlow et al.20 investigated the extraction of aqueous sodium hydroxide solution with organic solvent containing various quaternary ammonium salts by mixing sodium hydroxide with organic solvent containing R4NX. After phase separation, titration of the organic phase showed only traces of base presence if concentrated NaOH solution was employed and if Clwas the counter ion. The Cl-OH exchange was found to be of the order 1-2% for all quaternary ammonium chloride with chlorobenzene as solvent; i.e., 98% of the salts remained in the R4NCl form. However, upon addition of trace amounts of various alcohols, a dramatic change in the behavior of the system was observed and significant amounts of base could be detected in the organic phase. Table 13.3.3 shows the experimental results where 50% aqueous caustic solutions were extracted by equal volumes of 0.1 M (C8H17)4NBr in chlorobenzene containing 0.1 M of various alcohols.

806

Maw-Ling Wang

Table 13.3.3. Extraction of base by chlorobenzene solution of tetra-n-octyl-ammonium bromide and alcohols (0.1 M) from an equal volume of 50% NaOH (percent of the maximum possible basicity) % Primary alcohols

% 2-tert-Butylcyclohexanol

2.0

3- Methylcyclohexanol

2.0

Ethanol

4.5

1-Propanol

5.0

2-Methyl-1-propanol

4.4

tert-Butanol

0.3

1-Pentanol

4.3

2-Methyl-2-butanol

0.2

1- Hexanol

4.3

1-Heptanol

4.8

1,5-Pentanediol

< 0.02

1-Octanol

2.0

2,5-Hexanediol

5.2

1-Dodecanol

0.8

2,2-dimethyl-1,3-propanediol

18.4

2-Methyl-2,4-propanediol

28.0

Secondary alcohols

tert-Alcohols

Diols

2- Propanol

1.9

2,3-Dimethyl-2,3-butanediol

25.8

2-Pentanol

1.2

2,5-Dimethyl-2,5-hexanediol

32.0

2-Hexanol

1.1

2-Octanol

0.7

Ethylene glycol monoethylether

8.9

Cyclohexanol

0.5

Diethylene glycol monobutylether

8.7

4-tert-Butylcyclohexanol

1.5

Glycerol isopropylideneacetal

13.0

Diol monoethers

Data obtained from Dehmlow et al.20

It is apparent that the order of decreasing alkoxide extraction with monohydric alcohols is primary > secondary > tertiary. The better extractivity of diol anions can be attributed to the relatively high acidity of these alcohols in part, but it seems that the main factors are the distance between the two hydroxyl groups and the skeletal structure. In general, the concentration of the extracted base depends on the amount of alcohol added.20 The concentration of aqueous sodium hydroxide is also an important factor in the extraction processes. Herriott and Picker36 carried out the reaction of thiophenoxide ion with 1-bromooctane in a two-phase system. They found that an increase in the ionic strength of the aqueous phase or change to a more polar organic solvent increased the reaction rate. The effect of organic solvent on the reaction rate under NPTC is given in Table 13.3.4. Correlations between the rate constants and the partition coefficients indicate that the major function of the catalyst is simply the solubilization of the nucleophilic in the organic phase. Conventional methods of synthesizing ethers, i.e., Williamson synthesis and alkoxymercuration have been well developed in organic chemistry.76,96 The synthesis of formaldehyde acetal were carried out from the reaction of alcohol and dichloromethane in a 50% sodium hydroxide solution applying Tixoget VP clay as a catalyst. However, completing the reaction for such a low reaction rate takes long time. Dehmlow and Schmidt15 first


807

Table 13.3.4 Effect of solvent on the rate of used PTC technique to synthesize formreaction of thiophenoxide and bromooctane aldehyde acetals from alcohol and di-

chloromethane in the aqueous phase. Wang and Chang144-146 employed the PTC technique to synthesize formalde(C4H9)4N+IC7H16 0.02 hyde acetals from the reaction of alcohol C6H4Cl2 88 (C4H9)4N+Iand dibromomethane in an alkaline solution of KOH/organic solvent. Alcohol C6H4Cl2 0.45 (C3H7)4N+Br(ROH) first reacted with KOH so as to + 0.04 C6H5CH2(C2H5)3N Br C6H4Cl2 form potassium alkoxide (ROK) in the C6H4Cl2 28 C8H17(C2H5)3N+Braqueous phase. ROK further reacted with quaternary ammonium salt (QBr) C7H16 0.0093 (C6H5)4P+Brin the aqueous phase to produce quater+ C6H4Cl2 47 (C6H5)4P Br nary ammonium alkoxide (QOR) which C6H4Cl2 180 (C6H4)4P+Clis more soluble in the organic solvent. Dibromomethane reacted with QOR in the next step to form the desired product Data obtained from Herriott and Picker36 dialkoxymethane CH2(OR)2 in the organic phase, as shown in Equation [13.3.7]. Dibromomethane, which possesses weak dipole moment, may form a weak dipole-dipole bond with the organic solvent. However, this bond does not significantly affect the reaction rate. QOR solvates in a polar organic solvent. This solvation results in less energy in the nucleophilic agent than that in the transition state compound. The activation energy therefore becomes high due to the solvation of QOR with a highly polar Figure 13.3.1 Effects of the organic solvents on the conver- solvent which is unfavorable in the pression of CH2Br2; 6.88×10-2 mol of 1-butanol, 1.5 molar ratio ent reaction system. The low polarity of 1-butanol/1-octanol, 30 g of KOH, 2.76×10-2 mol of solvent neither solvates QOR, nor sepaCH2Br2, 3.11×10-3 mol of TBAB catalyst, 10 mL of water, 50 rates tetra-n-butylammonium ion (Q+) mL of organic solvent, 1020 rpm, 50oC (Adapted from Ref. from the alkoxide ion (-OR). Thus, the [145], by permission.) reactivity in low polar solvent is low. Organic solvents of appropriate polarity, such as chlorobenzene or dibutyl ether, are the best solvents to obtain a high yields of various alcohols, as shown in Table 13.3.5. As shown in Figure 13.3.1, similar results were obtained in synthesis of unsymmetric acetals under PTC conditions.144-146 The values of kapp, in which the reaction follows a pseudo-first-order kinetic rate law, are 4.59×10-3, 4.58×10-3, 8.17×10-3, and 1.47×10-2 min-1 for reaction of CH2Br2 with butanol and octanol in cyclohexane, n-decane, dibutyl ether and chlorobenzene, respectively. Catalyst

Solvent

k×103 M-1s-1

808

Maw-Ling Wang

Table 13.3.5. Effects of the organic solvents on the conversion of alcohols Organic solvent Reactant

chlorobenzene

dibutyl ether

xylene

benzene

Conversion, X, % 1-Butanol

83.22

58.94

49.65

55.48

1-Heptanol

81.44

50.16

41.94

44.90

1-Octanol

82.17

52.00

46.49

49.35

Cyclohexanol (2 h)

81.37

53.60

48.94

47.89

2-Ethoxyethanol (0.5 h)

92.82

87.86

79.61

73.86

2-(2-Ethoxyethoxy)-ethanol (0.5 h)

98.58

98.87

96.21

92.21

Dielectric constant

5.62

3.08

2.27

2.28

Data obtained from Wang and Chang;145 9.17×10-2 mol of alcohols, 10 mL of H2O, 0.028 mol of CH2Br2, 30 g of KOH, 1 g of TBAB catalyst, 50 mL of organic solvent, 1020 rpm, 50oC

(B) Synthesis of ether compound catalyzed by crown ether (NPTC) The other type of phase transfer catalyst is crown ether, cryptands, polyethylene glycol (PEG) and their derivatives, and other nonionic phase-transfer agents. The phase transfer agent complexes with inorganic cation, along with the anion, can be transferred to the organic phase, preparing for reaction with organic-phase reactant. For example: with 18-crown-6 ether as a phase transfer agent for sodium cyanide:

[13.3.8]

The function of crown ether is that it can chelate with metal ion, such as: lithium, sodium or potassium. Czech et al.12 noted that crown ethers are a better phase-transfer catalyst for solid-liquid reactions, whereas quaternary salts are better for a liquid-liquid system. Table 13.3.6 shows the solubilities of potassium salts in acetonitrile at 25oC in the presence and absence of 18-crown-6 ether. The solubility of potassium salts in CH3CN highly depends on the addition of 18-crown-6 ether.


809

Table 13.3.6. Solubilities of potassium salts in CH3CN at 25oC in the presence and absence of 18-crown-6 ether Potassium salt KF KCl KBr KI KCN KOAc KSCN

Solubility of potassium salt in 0.15 M crown in CH3CN (A) -3

4.30×10

-2

5.55×10

-1

1.35×10

-1

2.02×10

-1

1.29×10

-1

1.02×10

-1

8.50×10

in CH3OH (B)

Enhancement factor (A/B)

-4

13.52

-4

228.40

-1

64.90

-1

1.92

-3

108.40

-4

204

-1

1.13

3.18×10 2.43×10 2.08×10 1.05×10 1.19×10 5.00×10 7.55×10

Data adopted from the work of Liotta60

The rates of reaction of benzyl bro-

Table 13.3.7. Rates of reaction of benzyl mide and benzyl chloride with potassium chloride with potassium cyanide at 85oC in the presence and absence of 18-crown-6 cyanide were studied as a function of added water in the presence and absence of crown ether as a function of added water

ether in toluene at 85oC,61 as shown in Table 13.3.7. The reaction is highly affected by k×105 s-1 k×105 s-1 Water the addition of 18-crown-6 ether. In addi(crown) (no crown) tion to enhancing the reaction rate, it is im0.0 3.2 0.0 portant to note that in the absence of added 0.36 9.2 0.0 water, the rates followed zero-order kinetics, while in the presence of added water, 0.50 9.4 0.0 the rates followed first order kinetics. 1.00 11.6 0.0 (C) Synthesis of ether compound catalyzed by polyethylene glycols (NPTC) 2.00 14.7 0.0 Similar to quaternary ammonium 10.0 10.2 0.0 salts, polyethylene glycols (PEGs) act as 20.0 5.8 1.3 the phase transfer catalyst. There are two majors effects of PEG on the two-phase re40.0 3.9 1.9 actions. First, part of the PEG, existing in 75.0 4.8 3.2 the organic phase, forms a complex with metal cation. The formation of a complex Data obtained from the work of Liotta;61 0.05 mol of leads to an increase in the solubility of sobenzyl chloride, 0.01 mol of 18-crown-6 ether, 0.15 mol dium alkoxide (RONa) or sodium of KBr, and 0.015 mol of KCN phenoxide (PhONa) for the synthesis of ether in the organic phase. Hence, the reaction rate in the organic phase is promoted. Second, PEG acts as an excellent organic solvent, but it can also dissolve in water. Thus, part of the alkyl halide that is dissolved by PEG is brought into the aqueous phase from the organic phase. The dissolved alkyl halide directly

810

Maw-Ling Wang

reacts with phenoxide (PhO-) or alkoxide (RO-) ion in the aqueous phase, as shown in Figure 13.3.2.143 The reaction rate in the aqueous phase is also enhanced. The mechanism of the reaction rate of alkyl halide (or allyl halide, RX) and phenoxide (PhO-), both existing in the aqueous phase with PEG help, is different from that in presence of quaternary ammonium salt.

[13.3.9]

The reaction catalyzed by PEG can be carried out either in a homogeneous phase or in a two-phase solution. The alkyl halide usually serves as the reactant as well as the solvent. The reaction proceeds because organic-phase reactant dissolves in an organic solvent in the presence of PEGs. Dichloromethane, chlorobenzene, ethyl ether, cyclohexane and n-decane are frequently used solvents. The reaction mechanism of two-phase catalytic reaction by PEG includes formation of a com141,142 This Figure 13.3.2 Dependence of the concentration of allyl chloride in plex of PEG with cation. the aqueous phase on the amount of PEG-1500 added, Vorg=Vaq=50 is different than the reaction catamL, 30oC (Adapted from Ref. [143], by permission.) lyzed by quaternary salts. Table 13.3.8 shows the initial reaction rate using PEG, ((-r)i,PEG) and the initial reaction rate without using PEG, ((-r)i,B) in various organic solvents. Both (-r)i,B and (-r)i,PEG decrease when the polarity of the organic solvent increases. The maximum reaction rate is obtained with n-decane, which has the lowest polarity, as the protic solvent. Same results were obtained from the work of Landini et al.55 on the reaction of n-octylmethylene sulfonate and bromide ion in a homogeneous phase with C16H33P+(C4H9)3Y- as PTC. Wang and Chang142 made a reasonable explanation for this peculiar phenomena, i.e., the transition state possesses a higher degree of dispersity of electric charge than does the ground state. Increasing the polarity of the solvent increases the relative activation energy between the transition state and the reactants. Hence the reaction rate is decreased.


811

Table 13.3.8. Effect of the aprotic solvent on the initial reaction rate of the allylation of phenoxide

Solvents

Dielectric constant (20oC)

Initial reaction rate×103 Mh-1 (-r)i,B

(-r)i,PEG

(-r)i,PEG/(-r)i,B

n-decane

1.991

171.77

355.62

2.07

cyclohexane

2.023

127.5

277.68

2.16

ethyl ether

4.335

98

205.83

2.10

cyclobenzene

5.708

75.25

174.56

2.32

dichloromethane

9.080

45.61

86.31

1.89

Data obtained from the work of Wang and Chang142

13.3.1.3 Effects of the organic solvents on the reactions in other catalysts (A) Quaternary ammonium salts as NPTC The effects of the organic solvents on the reaction rate are given in Table 13.3.9.121 The rates of the reactions of the tetra-n-butylammonium and potassium salts of phenoxide with 1-chlorobutane and 1-bromobutane in pure solvents and solvent mixtures varying in dielectric constant from 2.2 to 39 were obtained by Uglestad et al.121 Table 13.3.9. Effect of organic solvent on the reaction of tetra-n-butylammonium salts of phenoxide and potassium salts of phenoxide with halobutane Rate constant×105, Lmol-1s-1 Reactants Solvents

1-C4H9Cl

1-C4H9Br

1-C4H9Br

K+-OC6H5

Bu4N+-OC6H5

K+-OC6H5

Bu4N+-OC6H5

0.01

330

2.2 6

Dioxane 10% CH3CN 50%

0.0025

2.8

0.22

400

CH3CN

0.084

4.0

12

600

0.33

2.2

40

300

Acetonitrile

Dielectric constant ( ε)

1-C4H9Cl

39

Data obtained from Uglestad et al.,121 0.2 M C6H5O-, 0.05 or 0.1 M C4H9X, 25oC

The rates of reaction of potassium phenoxide vary over three orders of magnitude with the changes in the dielectric constant of solvent, whereas the corresponding rates with the tetra-n-butylammonium salt vary by approximately a factor of 6. (B) Tertiary amines as NPTC Quaternary ammonium salts, PEGs and crown ethers are the common compounds, employed as PTC. The inexpensive tertiary amines have also been used as the phase transfer catalysts (PTC) in recent years. The synthetic process for producing 2mercaptobenzimidazole (MBI) is a reaction of o-phenylene diamine (C6H4(NH2)2) and carbon disulfide (CS2) in a two-phase medium affected by appropriate choice of solvent.128-130

812

Figure 13.3.3. Effect of solvents on the conversion of o-phenylene diamine in the two-phase catalyzed reaction; 3.17x10-3 mol of C6H4(NH2)2, 2.50×10-2 mol of CS2, 1.679×10-3 mol of tributylamine (TBA), 1000 rpm, 30oC. (Adapted from Ref. [128], by permission.)

Maw-Ling Wang

Figure 13.3.4. Effect of solvents on the conversion of o-phenylene diamine; 0.4 g of C6H4(NH2)2, 4.003 g of CS2, 0.4 mL of tributylamine (TBA), 50 mL of organic solvent, 600 rpm, 30oC. (Adapted from Ref. [129], by permission.)

Several solvents, such as: n-decane, n-hexane, benzene, chlorobenzene and dichloromethane, which are immiscible with water, were used. The effect of solvent on conversion is shown in Figure 13.3.3.128 The order of the conversion of o-phenylene diamine C6H4(NH2)2 in various organic solvents is: dichloromethane > chlorobenzene > benzene > n-hexane > n-decane, which is consistent with the order of solvent polarity, i.e., the greater the polarity of solvent the higher the conversion of o-phenylene Figure 13.3.5. Effect of solvents on the conversion of diamine. o-phenylene diamine; 3.18x10-3 mol of o-phenylene Results for the reaction of o-phenyldiamine, 8 molar ratio of CS2/C6H4(NH2)2, 0.01 M of triethylamine (TEA), and 50 mL of CH3CN, 600 rpm, ene diamine and carbon disulfide in a ho40oC. (Adapted from Ref. [130], by permission.) mogeneous phase (organic solvent) are given in Figure 13.3.4.129 The order of the reactivities is: dichloromethane (8.91) > chlorobenzene (5.6) > chloroform (4.8) > toluene (2.4) > benzene (2.3). The reaction rate is related to the dielectric constant of the organic solvent. A larger conversion of o-phenylene diamine was obtained using solvent with a higher dielectric constant. In choosing a polar organic solvent, such as: MeCN, MeOH, EtOH, DMSO, DMF and THF, a homogeneous solution was used for the reaction.130 Figure 13.3.5 shows the effects of organic solvents (protic or aprotic) on the conversion of o-phenylene diamine. The order of the reactivities for these six organic solvents is: DMF > DMSO > > MeCN > MeOH > EtOH >THF. The corresponding dielectric constants of solvents are: DMF (37.71), DMSO (46.45), MeCN (35.94), MeOH (32.66), EtOH (24.55) and THF (7.58), respectively. The protic solvents, such as MeOH and EtOH, containing hydroxyl group possess acidic proper-


813

ties. The unpaired electrons on the oxygen atom associate with the anions. A relatively lower conversion is obtained in MeOH or EtOH solvent. This result indicates that the acidic hydrogen bond does not have a strong catalytic capability. The aprotic solvents, which do not possess hydrogen bond, are highly polar. Therefore, the aprotic solvents possess high alkalinity and nucleophilicity required to obtain a high conversion of o-phenylene diamine in the synthesis of mercaptobenzimidazole (MBI). A larger conversion is obtained when using a protic solvent or aprotic solvent of high polarity. However, the structure of DMF, which is an amide, is similar to that the tertiary amine. It possesses similar catalytic property to dimethylaminopyridine (DMAP). The effect of DMF on the conversion of o-phenylene diamine is more pronounced than that of DMSO. The Arrhenius rate equations in various solvents for the reaction of o-phenylene diamine and carbon disulfide catalyzed by tributylamine are as follows: DMF:

kapp = 1.06x1015exp(-1.20x104/T)

DMSO:

kapp = 7.82x108exp(-7.78x103/T)

MeCN:

kapp = 1.39x1013exp(-1.09x104/T)

MeOH:

kapp = 9.62x1014exp(-1.24x104/T)

EtOH:

kapp = 3.84x1010exp(-9.29x103/T)

THF:

kapp = 3.25x1038exp(-2.99x104/T)

[13.3.10]

kapp is the apparent rate constant in which the reaction follows pseudo-first-order rate law. In two-phase phase-transfer catalytic reactions, the solvents significantly affect the reaction rate. The main reason is that the distribution of regenerating catalyst QX and the active catalyst QY between two-phases is highly dependent upon the polarity of the solvent. It is desirable for most of the intermediate products to stay in the organic phase and react with the organic-phase reactant. Therefore, a solvent with high polarity will be preferred for the reaction. (C) Pyridine 1-oxide (PNO) as IPTC The substitution reaction of benzoyl chloride (PhCOCl) and sodium acetate (CH3COONa) using pyridine 1-oxide (PNO) as the inverse phase-transfer catalyst (IPTC) in a two-phase system of organic solvent and water was investigated by Wang, Ou and Jwo.148-150 They found that the polarity of the organic solvent strongly affected conversion of benzoyl chloride, the yield of the main product (acetic benzoic anhydride (PhCOOCOCH3)), and the reaction rate. The reaction follows a pseudo-first-order kinetic rate law. Dichloromethane, chloroform, tetrachloromethane and cyclohexanone (C6H10O) were used as the organic modifier in the two-phase reaction system. The results are given in Table 13.3.10. A linear reaction rate was observed for a more polar organic solvent. The order of relative reactivities in these solvents is cyclohexanone > dichloromethane > chloroform > tetrachloromethane, consistent with their polarities. Kuo and Jwo54 obtained similar results. Wang, Ou and Jwo148 also found that the conversion was substantially increased with initial concentration of PNO increasing in the aqueous phase with CH2Cl2 present as

814

Maw-Ling Wang

the organic solvent. The reason is that the concentration of carboxylate ion54 influences the concentration of PNO in the organic phase. Table 13.3.10. Effect of the composition of organic solvent on the PNO-catalyzed PhCOCl-CH3COONa reaction in a two-phase medium

Organic phase

kapp×10-3 min-1 at T= 5oC

10oC

18oC

25oC

24.7

32.3

48.5(5.73)b

65.8(12.0)b

18.8

26.6

51.5 26.8 18.1

72.8

5.13

38.3 17.9 9.06

36.4(6.42)b

62.7(14.6)b

83.0

124

CH2Cl2+CHCl3 [CHCl3]=5.00 M

26.9

37.6

49.0

CH2Cl2+C6H10O [C6H10O]=3.00 M

35.0(4.49)b

57.2(9.38)b

91.4(20.4)b

14.6

19.5

26.7

CH2Cl2 CH2Cl2+CCl4 [CCl4]=1.00 M [CCl4]=3.00 M [CCl4]=5.00 M CH2Cl2+C6H5NO2 [C6H5NO2]=1.00 M

CHCl3

28.0(2.93)b

16.2

CCl4 C6H10O

33oC

68.7

106

170

Data obtained from Wang et al.148 2.00×10-4 M of PNO, 1.00×10-2 M of PhCOCl, 0.500 M of CH3COONa, 18oC, 1200 rpm, 50 mL of H2O, 50 mL of organic solvent, bNo PNO added; C6H10O, cyclohexanone

Figure 13.3.6. Effect of the mole fraction of CCl4 in the mixed organic solvent on the kapp value in the two-phase H2O/(CH2Cl2+CCl4) medium; 1.00×10-2 M of PhCOCl 0.500 M of CH3COONa, 2.00×10-4 M of PNO, 50 mL of H2O, 50 mL of organic solvent (CH2Cl2+CCl4), 18oC. (Adapted from Ref. [148], by permission.)

In the studies on the inverse phase-transfer catalytic reaction, Wang, Ou and Jwo148 conducted two independent experiments in order to evaluate the effect of polarity of the organic phase on the reaction. In the first experiment, a relatively inert organic substances such as C6H5CH2CN, C6H5N(Et)2, C6H5NO2, C6H5CN, CH3COOC2H5 or C3H7COOC2H5 were individually added to the organic phase (CH2Cl2) as the mixed organic solvent in the two-phase reaction system. The reactions of these compounds with PhCOCl or PNO were negligibly slow compared to the reaction of PhCOCl and PNO. The results are given in Tables 13.3.11 and 13.3.12, respectively. It is shown that kapp increased with added inert substance of high polarity,


Table 13.3.11. Effect of the inert organic substance on the PNO-catalyzed CH3COONa-PhCOCl reaction in a twophase H2O/CH2Cl2 medium Organic substance, R

kapp×103 min-1

Dipole moment, D

C6H5CH2CN

61.2

C6H5CN

60.5

C6H5N(Et)2

56.8

C6H5NO2

57.0

4.22

CH3COOC2H5

53.1

1.78

C3H7COOC2H5

48.5

CH2Cl2

48.5

1.60

CCl4

42.5

0

4.18

Data obtained from Wang et al.148 1.00×10-2 M of PhCOCl, 0.500 M of CH3COONa, 0.500 M of R, 2.00×10-4 M of PNO, 50 mL of H2O, 50 mL of CH2Cl2, 18oC

815

such as nitrobenzene and ethyl acetate or basic organic substance, such as diethylaniline. The kapp-value increased to a greater extent with added highly polar and basic organic substance, such as benzyl cyanide and benzonitrile. In the second set of experiments, reactions were carried out with nonpolar CCl4 added to the CH2Cl2 as the mixed organic solvent. As shown in Figure 13.3.6, that due to decreased polarity, the value of kapp also decreased with increased amount of added CCl4 to a minimum.148 Then, it increased slightly on further addition of CCl4 due to the increased rate of PNO-catalyzed hydrolysis of PhCOCl. Since the distribution of PhCOCl in the CH2Cl2 decreases with increased amount of CCl4, the reaction rate of PhCOCl with PNO in the aqueous phase leads to the hydrolysis of PhCOCl. Wang, Ou and Jwo148 observed that the yields of PhCOOCOCH3 decrease with increased content of CCl4.

Table 13.3.12. Effect of the amount of inert organic substance on the PNO-catalyzed CH3COONa-PhCOCl reaction in a two-phase H2O/CH2Cl2 medium kapp×103 min-1 with [R]org, M

Inert organic Substance [R]org

0.100

C6H5CH2CN

57.3

C6H5NO2

49.0

C6H5N(Et)2

55.1

CCl4

47.6

0.300

61.1

0.500

1.00

1.50

61.3

62.4

71.8

57.0

62.7

60.7

60.9

38.3

28.5

24.7

56.8 42.1

0.800

2.00

54.9

Data obtained from Wang et al.148 1.00×10-2 M of PhCOCl, 0.500 M of CH3COONa, 2.00×10-4 M of PNO, 50 mL of H2O, 50 mL of CH2Cl2, 18oC

In Table 13.3.12, kapp approached a constant value when nitrobenzene (1.0 M) was added. This result indicates that solvation of the transition structure for the reaction of benzoyl chloride with sodium nitrate reached an upper limit. Benzyl cyanide is a polar solvent. The value of kapp increased with increased content of benzyl cyanide. Further, highly basic diethylaniline (C6H5N(Et)2) could increase the concentration of free PNO and also the reaction rate. However, this compound is less polar than dichloromethane and the polarity decreased with increased proportion of diethylaniline, which caused the value of kapp to de-

816

Maw-Ling Wang

crease. Therefore, the value of kapp reached a maximum, as shown in Table 13.3.12. In general, the value of kapp increased with an increased proportion of highly polar inert organic substance, such as C6H5CH2CN and C6H5NO2, and decreased with increased proportion of slightly polar inert organic substance, such as C6H5N(Et)2 and CCl4. In the case of CCl4, the greater the proportion of nonpolar or less polar compound, the smaller the reaction rate. The results are due to a combination of the hydrolysis of benzoyl chloride, the distribution of PNO between two phases, and the mass transfer of PNO from organic phase to aqueous phase. (D) Electrophile reaction by Table 13.3.13. Initial rates of coupling between NaDBS 4-nitrobenzene diazonium chloride and Although the electrophile N-ethylcarbazole at 0oC in various solvent mixtures transferred to the organic phase in the presence and absence of additives from the aqueous phase by phase transfer catalysis (PTC), the role c 9 Rate ×10 , of organic solvent was still imporAdditveb Solventa -1 -3 mol dm tant. Ellwood and Griffiths24 carCH2Cl2-H2O none 1.25 ried out the coupling reactions between 4-nitrobenzendiazonium NaDBS 11.1 CH2Cl2-H2O chloride and N-ethylearbazole or 18-crown-C-6 ether 0.28 CH2Cl2-H2O N,N-diphenylamine in aqueous Lissapol NXd 0.56 CH2Cl2-H2O media. The coupling reactions were accelerated by using a none < 0.1 Toluene-H2O two-phase water-dichloroNaDBS < 0.1 Toluene-H2O methane containing sodium 4-dodecylbenzene sulfonate none 0.89 EtOAc-H2O (NaDBS) as a transfer catalyst for NaDBS 1.03 EtOAc-H2O the diazonium ion. Effects of solnone 1.39 PhNO2-H2O vents and catalyst (NaDBS) on the rate constants are given in TaNaDBS 10.5 PhNO2-H2O ble 13.3.13. The NaDBS (0.05 none 0.47f AcOH-H2Oe molar proportions) increases the DMF-H2Og none 0.53f rate coupling in dichloroh methane-water by a factor of at 1,4-Dioxane-H2O none 1.33 least 20 relative to the reaction in water-acetic acid. A part of this 24 a Data obtained form the work of Ellwood and Griffths; v/v=1/1 except where stated, b0.05 mmol except where stated, cinitial rate of increase is attributed to incomformation of azo dye, dcommercial (ICI) non-ionic detergent, 0.05 plete solubility of N-ethyl mmol, e95% H2O v/v, containing NaOAc·3H2O, fsuspension of carbazole in the latter solvent. g h N-ethylcarbazole, 60% H2O v/v, 60% H2O v/v homogeneous soluAlso, the polarity of the organic tion containing 1.0 mmol diazonium ion and 1.0 mmol N-ethyl phase is important (cf. CH2Cl2, carbazole 100 mL solvent C6H5NO2 have much higher dielectric constants than toluene and ethyl acetate). This may be attributed to the covalent character of diazonium arylsulfonates. Its ionization is greater in the former solvents. Crown ethers act as transfer agents for the diazonium ion, but the resultant complexes have low coupling reactivity.


817

(E) Oxidation by dimethyl polyethylene glycol and oxidant Dimethyl polyethylene glycol solubilizes potassium permanganate in benzene or dichloromethane and can thus be used as a phase-transfer agent for permanganate oxidation. The reaction is highly dependent on the organic solvent. If benzene is used as solvent, dimethyl polyethylene glycol does not efficiently extract KMnO4 from an aqueous solution, but it solubilizes the solid reagent when CH2Cl2 is used as the solvent, KMnO4 may be transferred from either aqueous solution or from the solid phase. The effect of organic solvent on the distribution of products is given in Table 13.3.14.58 Table 13.3.14. Oxidation of cyclododecene Phase transfer agent

Solvent system

Oxidation a ratio

Benzene+17% acetic acid

3.3

Polyetherb

1,2-Cyclododecanedione (16), dodecanedioic acid (59), cyclododecane (23)


3.3

Crown etherc



3.3

Adogen 464


Dichloromethane+17% acetic acid

3.3

Polyether



3.3

Crown ether

1,2-Cyclododecanedione (7), dodecanedioic acid (83), cyclododecene


3.3

Adogen 464

1,2-Cyclododecanedione (7), dodecanedioic acid (83), cyclododecene (2)


2.2

Adogen

1,2-Cyclododecanedione (18), dodecanedioic acid (63), 2-hydroxycyclododecanone (6)


1.6

Agogen 464

1,2-Cyclododecanedione (14), dodecanedioic acid (40), 2-hydroxycyclododecanone (6), cyclododecene (23)


2.2

Adogen 464

1,2-Cyclododecanedione (19), dodecanedioic acid (27), 2-hydroxycyclododecone (7), cyclododecene (23)


2.2

Adogen 464

1,2-Cyclododecanedione (69), dodecanedioic acid (13), 2-hydroxycyclododecane (3), cyclododecene (9)

Dichloromethane/water +10% acetic acid

2.2

Polyether

1,2-Cyclododecanedione (16), dodecanedioic acid (82)

Dichloromethane/aqueous NaOH

1.0

Benzyl triethyl ammonium chloride

1,2-Cyclododecanediol (50)

Products (%)

Data obtained from Lee and Chang;58 anumber of moles of potassium permanganate per mole of alkene, b dimethylpolyethylene glycol, cdicyclohexano-18-crown-6 ether

818

Maw-Ling Wang

(F) Polymerization by PTC Poly(ethylene glycol)-block-poly(butylacrylate), synthesized by radical polymerization,56 were obtained by PTC in the Williamson reaction. The morphology and the crystallinity of the cast film of the block polymer were significantly affected by the organic solvent. The technique of phase-transfer catalysis has been extensively applied to the two-phase polycondensation using various phase-transfer catalysts, such as quaternary ammonium and phosphonium salts, crown ethers and poly(ethylene glycol)s.8,11,30,46,53,75,87,119,151 Various types of condensation polymers such as aromatic polysulfonates and polysulfides, aromatic polyethers, aliphatic and aromatic polysulfides, and carbon-carbon chain polymers of high molecular weights by the phase-transfer catalyzed polycondensation from combinations of aromatic disulfonyl chlorides, phosphonic dichlorides, activated aromatic dichlorides, and aliphatic dihalides, with bisphenol, aliphatic and aromatic dithiols, and active ethylene compounds. The two-phase polycondensation was generally carried out in a water-immiscible organic solvent-aqueous alkaline solution system at room temperature. The method of polycondensation offers a highly versatile and convenient synthetic method for a variety of condensation polymers. Aromatic polysulfonates of high molecular weights can be prepared from aromatic disulfonyl chlorides and alkaline salts of bisphenols by interfacial polycondensation technique using onium salt accelerators. In the absence of the catalyst, only low molecular polysulfonate III was obtained, even though the reaction was continued for 254 hours, whereas the addition of these quaternary ammonium salts and crown ethers increased the average molecular weight of the polymer remarkably.

[13.3.11]

Among the catalysts, TBAC and DC-18-C-6 were found to be highly efficient, leading to the formation of the polysulfonate with an inherent viscosity [η] of as high as 1.4 dLg-1. Similarly, aromatic polysulfonates and aromatic polyether were synthesized from the polycondensation of phenylphosphonic dichloride (IV) with bisphenol A (II) leading to a polyphosphonate (V), aromatic dihalides (VI) with alkaline salts of bisphenols (VII), under various phase transfer catalysis in two-phase system, i.e.,

[13.3.12]


819

[13.3.13]

The effect of organic solvent on the two-phase polycondensation is shown in Table 13.3.15. Chloroform, 1,2-dichloroethane, nitrobenzene, acetophenone and anisole were all effective as the polymerization media to produce moderate molecular weight polymer (VII). Table 13.3.15. Synthesis of aromatic polyether VII in various organic solvent/water system with DC-18-C-6 catalyst Reaction temp., oC

Reaction time, h

Polymer [η], dLg-1*

CH2Cl2

20

24

0.84

CHCl3

20

24

0.53

CH2ClCH2Cl

20

24

0.42

C6H5NO2

20

24

0.47

C6H5NO2

80

2

0.51

C6H5NO2

100

1

0.42

Solvent

Data obtained from the work of Imai;41 Reaction conditions: 2.5 mmol of II, 2.5 mmol of VI, 0.05 mmol of DC-18-C-6 in 3.5 mL of solvent, and 5 mL of KOH (1.01 M) solution, *Measured at a concentration of 0.5 dLg-1 in DMF at 30oC

A convenient method for the preparation of polysulfides by the two-phase polycondensation in a KOH solution is known.41 Polycondensation of 1,4-dibromobutane (VIII) and 1,6-hexanedithiol (IX) leading to polysulfide (X) was carried out in various organic solvent/H2O system with DC-18-C-6 catalyst.

[

]

Br (CH2 ) 4 Br + HS(CH2 ) 6 SH → −(CH2 ) 4 − S − (CH2 ) 6 − S − VIII IX X

n

[13.3.14]

The results of the polycondensation are given in Table 13.3.16. All polymerization media employed produced polysulfide with moderately high inherent viscosities; whereas the polymer with the highest viscosity was produced in the absence of organic solvents. Polycondensation conducted in the presence of any catalyst in this system led to the formation of a polymer with moderately high molecular weight.

820

Maw-Ling Wang

Table 13.3.16. Synthesis of aliphatic polysulfide Several polycarbonates59 were X in various organic solvent-water system with synthesized by two-phase DC-18-C-6 catalyst polycondensation of bisphenols and

brominated with trichloromethyl chloroformate in a system of an organic solvent and aqueous alkaline 20 0.30 CH2Cl2 solution of quaternary ammonium 80 0.31 CHCl3 salts. Chlorinated hydrocarbons, di80 0.30 C6H6 chloromethane (DCM), tetrachloromethane (TCM), tetrachloromethane 80 0.58 C6H5NO2 (TCM) and nitrobenzene (NB) 80 0.58 CH3CN served as organic solvents. The effects of solvents on the reaction None 80 0.73 yields are given in Table 13.3.17.59 41 Although polymers with a high yield Data obtained from Imai; Polymerization conditions: 2.5 mmol of VIII and IX, 0.05 mmol of DC-18-C-6 in 2.5 mL of solvent and were obtained using nitrobenzene 5 mL of 1.01 M KOH solution for 48 h, *measured at a concentra- (NB) as an organic solvent, the intion of 0.5 gdL-1 in chloroform at 30oC herent viscosities were low. Polycarbonates were prepared by a two-phase condensation of TCF with bisphenol S. They precipitate from chlorinated hydrocarbon solvents such as DCM, TCM and DCE. According to both the yield and the inherent viscosity of these polymers, the use of BTEAC as a phase-transfer catalyst, sodium hydroxide as a base and DCE as an organic solvent was suitable to prepare a polycondensate having a large molar mass and a high yield. Reaction temp. oC Polymer [η], dLg-1*

Solvent

Table 13.3.17 Synthesis of bisphenol S-based homopolycarbonate by two-phase polycondensation catalyzed by PTCa Reaction conditions b

Polymer yield

Solvent

Catalyst

%

[η], dLg-1*

State

DCM

TBAB

76.7

0.21

ppt.

DCM

TBAC

76.6

0.10

ppt.

DCM

BTEAC

78.7

0.13

ppt.

DCM

BTEAB

79.3

0.20

ppt.

TCM

TBAB

60.0

0.11

ppt.

TCM

BTEAC

83.4

0.12

ppt.

DCE

TBAB

88.9

0.28

ppt.

DCE

TBAC

76.4

0.12

ppt.

DCE

BTEAC

86.5

0.32

ppt.

DCE

BTEAB

89.3

0.21

ppt.

NB

TBAB

85.0

0.17

solution

NB

TBAC

90.6

0.11

solution


Reaction conditions b

821

Polymer yield

Solvent

Catalyst

%

[η], dLg-1*

State

NB

BTEAC

92.2

0.19

solution

Data obtained from Liaw and Chang;59 aPolymerization was carried out with bisphenol S (5.00 mmol) and TCF (7.50 mmol) in the organic solvent (37.5 mL) and water (30 mL) in the presence of catalyst (3.15 mmol) and sodium hydroxide (28.5 mmol) at room temperature for 2 h. bAbbreviations: DCM, dichloromethane; TCM, tetrachloromethane; DCE, 1,2-dichloroethane; NB, nitrobenzene. *Measured at a concentration of 0.5 gdL-1 in DMF at 25oC

Another type of polysulfide (XIII) was synthesized by the two-phase polycondensation of bis-(3-chloroacryloy)benzenes (XIa and XIb) with 4,4’-oxybisbenzenethiol (XII). The polycondensation was carried out in a chloroform-water system at room temperature with some phase transfer catalysts.

[13.3.15]

Table 13.3.18 shows the results of polycondensation. The polysulfides having inherent viscosities above 0.5 dLg-1 were readily obtained from two bis(2-chloroacryloyl)benzene with or without use of phase transfer catalysts. These activated dichlorides are highly reactive, almost comparable to ordinary dicarboxylic acid chlorides. The use of catalysts, such as DC-18-C-6, was not essential to this type of polycondensation for producing high molecular weight of polysulfides XIII. Table 13.3.18. Synthesis of polysulfides XIII in organic solvent-water systema Dichloride

Solvent

Catalyst

Reaction time, min

Polymer, [η], dLg-1*

XIa

chloroform

none

15

0.21

XIa

chloroform

none

60

0.61

XIa

chloroform

DC-18-C-6

10

0.62

XIa

chloroform

TBAC

60

0.72

XIa

dichloromethane

DC-18-C-6

60

0.55

XIb

chloroform

none

60

0.55

XIb

dichloromethane

none

15

0.42

XIb

dichloromethane

DC-18-C-6

15

0.51

Data obtained from Imai;41 aReaction conditions: 2.5 mmol of XI, 2.5 mmol of XII, 0.05 mmol of catalyst, 5 mL of solvent, 5 mL of 1.01 M KOH at 15oC under nitrogen, *Measured at a concentration of 0.5 gdL-1 in concentrate sulfuric acid at 30oC

822

Maw-Ling Wang

13.3.1.4 Effect of the volume of organic solvent and water on the reactions in various reaction systems In general, the concentration of reactants in the aqueous phase is decreased by increased amount of water. The addition of water probably also decreases the concentration of the intermediate product in the organic phase. Hence, both the mass transfer rate and the degree of hydration with the anion are decreased, which also decreases the reaction rate. However, this argument is not necessarily correct. Figure 13.3.7 indicates that the conversion in the two-phase reaction is not affected by the amount of water added.147 Wang and Yang135 studied the effects of the volume ratio of water to chloroFigure 13.3.7. Effect of the amount of water on the appar- benzene on the conversion for the reaction ent rate constant (kapp); 4 g of 2,4,6-tribromophenol, 0.9 g of KOH, 0.6 mL of benzyl bromide, 50 mL of H2O, 50 of 4-bromophenol and allyl bromide in an mL of chlorobenzene, 40oC. (Adapted from Ref. [147], alkaline solution of KOH/chlorobenzene at by permission.) 50oC under phase-transfer catalytic conditions. The reaction followed the pseudo-first-order rate law and the corresponding apparent rate constant decreased gradually when the water content was increased, as shown in Figure 13.3.8.135 The reason was that the concentration of the intermediate product tetra-n-butylammonium phenoxide (ArOQ, or the active catalyst) in the aqueous phase decreased with the increase in the amount of water. The mass transfer rate of the intermediate product (or the active catalyst) from the aqueous phase to the organic Figure 13.3.8. Effect of the volume ratio of water to chlorobenzene phase decreased when a large on the conversion; 1.568 g of 4-bromophenol, 1.0 g of KOH, 0.7 g of amount of water was used. In adallyl bromide, 0.2 g of TBAB catalyst, 50 mL of chlorobenzene, dition, the dilution effect led to re50oC. (Adapted from Ref. [135], by permission.) duction of the reaction rate in the aqueous phase. In general, a higher concentration of the intermediate product (tetra-n-butylammonium alkoxide, or the active catalyst, ArOQ) in the aqueous phase enhances the reaction rate. This is due to a large concentration gradient across the interface in transferring the species from the aqueous phase to the organic phase. For the reaction of allyl bromide and 2,4-dibromophenol in synthesizing 2,4-dibromophenyl allyl ether in an


823

alkaline solution of KOH/chlorobenzene two-phase medium under PTC conditions,132 the conversion increases with the increase in the concentration of ArOQ in the aqueous phase (or decreasing content of water). However, this change is small, reflecting a small mass transfer resistance, as shown in Figure 13.3.9.132 The influence of the amount of water on the conversion in the reaction of carbon disulfide and o-phenylene diamine catalyzed by tertiary amine in a two-phase medium was studied. The conversion decreased with the increase in the Figure 13.3.9. Effect of the water content on the conversion; 2.28 g amount of water. Therefore, the of 2,4-dibromophenol, 0.2 g of TBAB catalyst, 0.7 g of allyl brovalue of apparent rate constant mide, 1.0 g of KOH 50 mL of chlorobenzene, 50oC. (Adapted from (kapp), in which the reaction folRef. [132], by permission.) lows pseudo-first-order-rate law, decreases with the increase in the volume of water.128 Wang and Chang144-146 found that the conversion increases with the increase in the volume ratio of water to organic solution up to 1/5. The conversion is independent of the volume ratio of water to organic phase (chlorobenzene), greater than 1/5. The reason is that the reaction is carried out in a large amount of KOH (solid form). Probably, the omega phase is generated for the volume ratio of water to chlorobenzene less than 1/5. However, this change in the conversion vs. the volume ratio of water to chlorobenzene is not significant. (A) Omega phase reaction It is found that 92% of the 18-crown-6 ether added to a salt (KCN and KCl) and toluene system resided in the organic phase. However, all but approximately 1-2% of the crown ether was translocated onto the surface of the salt upon addition of small quantities of water. The results of Liotta et al.62 are given in Table 13.3.19. The initial water added to the system coats the surface of the salt particles and it was this aqueous salt coating that extracted the crown from the organic phase. Liotta et al.61,62 called this new region of the reaction system the omega phase. The 8% of the crown located on the surface of the salt particles prior to the addition of water was probably due to the presence of water already present in the salt. The distribution of 18-crown-6 ether between the organic phase and the omega phase was determined.62 The amount of crown ether in the organic phase remained low and relatively constant (0.06-0.07 milimoles of 18-crown-6 ether in organic phase). The omega phase adsorbed most of added crown ether. For the accompanying pseudo first-order kinetics reaction of benzyl bromide with potassium cyanide, the results are given in Table 13.3.20.122 There is a slight increase in the rate as the number of millimoles of 18-crown-6 ether increases, but the rate remains essentially constant with the increase of crown ether.

824

Maw-Ling Wang

Table 13.3.19 Effect of added water on the concentration of 18-crown-6 ether in toluene at room temperatures Water, µL

Equiv. of water, mole of H2O/mole of crown

Percent crown in toluene, %

0

0.00

91.5

10

0.14

81.4

15

0.21

77.3

22

0.31

50.0

25

0.35

34.6

30

0.42

17.7

45

1.25

2.5

50

1.39

2.0

80

2.22

1.0

Table 13.3.20 18-Crown-6 catalyzed reactions of benzyl bromide with KCN as a function of added crown ether Millimoles of 18-crown-6

k×105 sec-1

3.0

2.16, 2.47

5.0

3.97, 3.63

7.0

3.86, 3.99

10.0

3.75, 4.00

12.0

3.80, 3.60

Data obtained from Vladea and Simandan;122 1.0 mL of H2O, 0.15 mole KBr, 0.15 mol of KCN, 50 mL of toluene, 25oC

Data obtained from Liotta et al.62 0.0040 mole of 18-crown-6, 0.027 mol of KCN, 10 mL of toluene

(B) Reaction catalyzed by PEGs The structure of polyethylene glycol (HO(CH2CH2O)nH, PEG) is similar to that of crown ether. Polyethylene oxide chains (CH2CH2O) form complexes with cations, much like crown ethers, and these complexes cause the anion to be transferred into the organic phase and to be activated.122 Table 13.3.21 shows the binding constant, K, for PEG complexes with sodium cation depend on both the value of n (i.e., average molecular weight of PEG or number of (CH2CH2O) unit) and on the end-group substituents.112,114,115 Gokel and coworkers31,32 determined the binding strength for Na+ in anhydrous methanol solution with PEGs and obtained the binding constant K=1.4. They concluded that the strength of complexation is a function of the total number of binding sites present and not the number of polymer chains, suggesting that a long PEG chain may be involved in binding more than one cation. PEGs and their derivatives have been extensively investigated as phase transfer catalysts and are used in many commercial processes. In the absence of strong acids, PEGs are nontoxic, inexpensive, and thermally stable. For some reactions such as with hydroxide transfer, PEGs are excellent catalysts, sometimes better than crown ethers, especially when used in liquid-solid PTC reactions with potassium salts, and with little or no added water, and with at least moderately polar organic solutions. PEGs are water soluble and if the organic phase is not sufficiently polar the PEG will reside almost completely in the aqueous phase; or with concentrated aqueous solutions of organic salts, the PEG may form a third catalyst-rich phase, a change that normally leads to a high level of catalytic activity. PEGs are themselves soluble in water. To obtain partitioning of PEG into an organic solution may require use of a mono- or diether derivative. Harris and Case34 found that with


825

exception of dichloromethane as an organic-phase solvent, most PEGs are themselves partitioned into the aqueous phase, depending also on the concentration of dissolved salts in the aqueous phase.35 To improve organic-phase solubility of PEG, several dialkyl ethers of PEGs as permanganate-PTC catalysts are suggested in Table 13.3.22. Table 13.3.21 Binding constants for complexation of PEGs and some symmetrical derivatives with sodium cation Na+ + PEG = [Na+⋅PEG] complex Log K (binding constant) with various sodium salts

PEG Avg. MW

Avg. n

HO-

200

4.1

1.64

300

6.4

2.02

400

8.7

2.26

600

13.2

2.59

2.09

1.99

1.87

1000

22.3

2.88

2.55

2.48

2.37

1500

33.7

3.09

2.86

2.80

2.68

2000

45.0

3.28

3.08

3.05

2.81

CH3O-

C2H5O-

PhO-

cyC5H10N-

0.5 1.55

1.25

1.05

1.16

1.49

1.51

2.46

3.08

Data obtained from Szabo et al.112,114,115

Aliphatic hydrocarbons are im-

Table 13.3.22 Partition of PEG-dialkyl ethers miscible with PEGs. Therefore, it is between C6H6 and H2O

important to select a good organic-phase solvent such as aromatic Partitioning, % in hydrocarbons, chlorinated hydrocarPEG Ether C6H6/% in H2O bons, or acetonitrile.80 In toluene, 14 C4-PEG1500-C4 PEGs are more effective catalysts than crown ethers for the reaction of benzyl 84 C6-PEG1500-C6 chloride and solid potassium acetate. Emulsion C18-PEG6000-C18 In butanol, the effectiveness of PEGs 108 C18-PEG750-Me and crown ethers as phase-transfer catalysts were the same for the reaction of 39 C18-PEG1900-Me benzyl chloride and solid potassium 37 C18-PEG5000-Me acetate.27 12 C8-PEG5000-Me 13.3.1.5 Effects of organic solvents on other 13 C4-PEG5000-Me phase-transfer catalytic PEG6000 chloroform > toluene > cyclohexane. The selectivity was 100% except when toluene was used as organic solvent. For this case, the concentration of cyclohexanol in carbon tetrachloride is higher than that of the other solvents.

828

Maw-Ling Wang

Table 13.3.24. Effect of organic solvent on the conversion and selectivity Solvent

Conversion, %

Selectivity, %

C.E., %

Energy consumption, KWH/mole

CCl4

42

100

83

0.226

CHCl3

38

100

70

0.268

Toluene

17

81

34

3.154

Cyclohexane

12

100

23

4.662

Data obtained from Tasi and Chou;120 Reaction conditions: 0.8 M of cyclohexanol, 15 mAcm-2 of current density, graphite cathode and graphite anode, 0.007 M of RuO2, pH=4, 0.99 Fmol-1 electricity passed, NaCl saturated solution as electrolyte, 1275 rpm, 5oC

13.3.1.6 Other effects on the phase-transfer catalytic reactions Simple mechanical separation such as filtration, centrifugation or phase separation can be used to separate the product and the phase-transfer catalyst by use of insoluble catalysts. However, the more frequently encountered technical problems in use of PTC for industrial applications is the need to separate the product and the phase-transfer catalyst by chemical equilibrium separation method in the liquid-liquid two-phase phase transfer catalytic reaction. The most commonly used methods for separation of products and PTC catalysts on an industrial scale are extraction and distillation. Other separation methods include sorption3,33,57 and reaction.45 The principle of extraction method used to separate PTC and product is based on solubility of quaternary ammonium salt in alkaline aqueous solution.2,25,104 For example, tetrabutylammonium bromide is soluble to the extent of 27% in dilute (1% NaOH) aqueous solutions, but when the solution is made more concentrated (15% NaOH), the solubility of Bu4N+Br- decreases to 0.07%. When the products are obtained in PTC system, they can be usually separated from PTC by distillation method. PTC catalyst in the distillation residue may sometimes be reusable. With quaternary ammonium salts as catalysts, temperatures above 100-120oC usually result in partial or total decomposition of the quaternary salts to trialkylamines and other products. Mieczynska et al.70 and Monflier et al.72 investigated the hydrogenation and hydroformylation under phase transfer catalytic conditions. They found that the yield of aldehydes obtained in hydroformylation of 1-hexene strongly depends on solvent: 24% in toluene, 53-86% in toluene-water-ethanol mixture and 77-94% in water-ethanol solution. The mixture of water-ethanol as a solvent was also found to be the best for hydrogenation of 1-hexene (96% of hexane). Conversion of Ph2PCH(CH3)(COOH) phosphine into sodium salt Ph2PCH(CH3)(COONa) yields aldehyde in toluene, 92% in toluene-water and 94% in toluene-water-ethanol mixture. In principle, hydroxide anion is very difficult to transfer from aqueous to organic phases, yet it is one of the most valuable and most commonly used anions in the PTC systems. Addition of small amounts of alcohols to PTC systems requiring hydroxide transfer causes a dramatic increase in rates. Therefore, addition of alcohol enhances the PTC reaction as the cocatalytic effect. For example: formation of alkoxide anions, RO-, which are more readily transferred than the highly hydrated hydroxide anion, and which can serve as a strong base just as well as OH-, and solvation of the hydroxide with alcohol rather than with water, making the hydroxide anion more organophilic and more easily transferred.99,100


829

Cyanide displacements catalyzed by quaternary ammonium salts usually do not proceed without the presence of water to facilitate exchange and transfer of anions. However, PTC displacement depends on alcohol structure. Benzyl alcohol is about 1.5-2 times as effective as either methanol or ethanol. In the synthesis of BTPPC (benzyltriphenylphosphonium chloride) from benzyl chloride and triphenylphosphine, second-order rate constants and activation parameters for the reaction of benzyl chloride and triphenylphosphine were measured in several protic and aprotic solvents covering a wide range of dielectric constant were obtained by Maccarone et al.64 Wang, Liu and Jwo127 also used eight solvents in studying their effect on the reaction of triphenylphosphine and benzyl chloride. They classified these solvents into two categories depending on the solubility of benzyltriphenylphosphonium chloride (BTPPC). Solvents that dissolve BTPPC are acetic acid, dichloromethane, methanol and water. Solvents that do not dissolve BTPPC are acetone, benzene, toluene and ether. In general, triphenylphosphine (TP) does not dissolve in methanol or water. The effect of solvents on the reaction rate was measured by the apparent rate constant in which the reaction follows pseudo-first-order rate law. The order of relative activities of solvents is methanol (0.34 h-1) > acetic acid (0.176 h-1) > dichloromethane (0.0468 h-1) > acetone (0.0114 h-1) > diethyl ether (0.0043 h-1) > benzene (0.0018 h-1) > toluene (0.0008 h-1). Table 13.3.25. Second-order rate constants and activation parameters for the reaction of benzyl chloride with triphenylphosphine in various solvents k×104, Lmol-1s-1


60oC

Decalin

2.26

0.00134

Toluene

2.38

0.0169

0.0843

0.181

0.353

Anisole

4.33

0.0569

0.260

0.466

1.15

Bromobenzene

5.40

0.0933

0.276

0.457

0.909

Chlorobenzene

5.62

0.0512

0.243

1.13

Benzyl alcohol

13.1

8.94

38.7

107

1-Butanol

17.1

4.93

19.5

43.7

Acetopnenone

17.39

0.300

1.32

9.19

1-Propanol

20.1

4.41

9.86

Acetone

20.3

0.166

0.417

Ethanol

24.3

3.60

7.73

Benzonitrile

25.2

0.545

1.71

6.38

Nitroethane

28.06

0.470

3.05

12.9

Methanol

32.65

8.86

N,N-dimethylformanide

36.7

0.460

1.88

6.59

Acetonitrile

37.5

1.20

Solvent

70oC

80oC

90oC

100oC 0.0355

2.79

830

Maw-Ling Wang

k×104, Lmol-1s-1

Solvent


60oC

N,N-Dimethylacetamide

37.8

0.29

1.06

4.29

N-Methylformamide

189.5

6.08

28.7

86.5

70oC

80oC

90oC

100oC

Data obtained from the work of Maccarone et al.64

The BC-TP reaction shows better reactivity in protic or polar solvent since the activated complex is more polar than both reactant molecules. 13.3.2 THREE-PHASE REACTIONS (TRIPHASE CATALYSIS) As stated, the solid PTC is suitable for the industrial processes concerning the removal of the catalyst from the reaction mixture and its economic recycle. The real mechanism of reaction in a triphase catalysis is not completely understood. However, the reaction rate and the conversion of reactant in a triphase catalysis (TC) is highly dependent on the organiphilicity (hydrophilicity or hydroprobicity) of the polymer support of the catalyst and the polarity of the organic solvent. Not only the partition of the organic to the aqueous solutions is affected by the organophilicity of the polymer-supported catalyst, but also the concentration distribution of the catalyst between two phases is influenced by the organophilicity of the polymer-supported catalyst. Ohtani et al.82-86 used polystyrene-supported ammonium fluoride as a phase transfer catalyst (triphase catalysis) for several base-catalyzed reactions, such as cyanoethylation, Knoevenage reaction, Claisen condensation and Michael addition. The catalytic activity of the polystyrene-supported ammonium fluid was comparable to that of tetrabutylammonium fluoride (TBAF). The ionic loading and the ammonium structure of the fluoride polymers hardly affected the catalytic efficiency. The reaction was fast in a non-polar solvent (e.g., octane or toluene) from which the rate-determining step of the base-catalyzed reaction is very similar to that of the SN2 nucleophilic substitution reactions. The solvent may affect the catalytic activity in several ways. The greater its swelling power, the larger the volume fraction of catalytic occupied by the more mobile liquid, and the swollen volume fraction of the more rigid polymer network. The degree of swelling and the viscosity within the polymer matrix affect intraparticle diffusion rates. The solvent may also affect intrinsic reactivity at the active sites. Experimentally, it is difficult to distinguish solvent effect on diffusivity from solvent effects on reactivity. Tomoi and Ford116 found that the triphase catalysis followed pseudo-first-order rate law. The corresponding apparent rate constant kapp decreases with solvent in the order: chlorobenzene > toluene > decane over wide ranges of particle sizes and polymer crosslinking. The ability of the solvent to swell the catalysts decreases in the same order. 13.3.2.1 The interaction between solid polymer (hydrophilicity) and the organic solvents In triphase catalysis, solvated resin supports are important carriers for solid-phase organic synthesis in combinatorial chemistry. The physical properties of resin, resin swelling, dynamic solvation, and solvated supports are important factors in affecting the synthesis.160 However, these factors are also affected by solvent. Selective solvation of resin alters the local reactivity and accessibility of the bound substrate and the mobility of the entrapped re-


831

agent. Resin solvation changes during the course of the reaction when the attached substrate changes its polarity or other physicochemical properties. The basic steps involved in reactions with resin-supported PTC catalysts differ from ordinary two-phase PTC reactions in one important respect: ordinary PTC reactions require only one reagent to be transferred from their normal phase to the phase of the second reactant. Use of resin-supported catalysts requires that both reagents diffuse to active PTC sites on the catalyst surface, or for reactions with slow intrinsic rates, both reagents must also diffuse to the active sites inside the resin bulk phase. The need for diffusion processes with solid catalysts also means that both reagents are required to diffuse to and penetrate the stagnant outer layer of liquid(s) (the Nernst layer), coating the catalyst particle. Ford and Tomoi28 carried out the reaction of 1-bromooctane with aqueous sodium cyanide, catalyzed by tributylphosphonium groups bound into beads of an insoluble styrene-divinylbenzene resin, C 8 H 5 PBu 3 Br 1− C8H17Br( org ) + NaCN ( aq )  polymer  −  → 1− C8H17CH( org ) + NaBr( aq ) [13.3.16]

The reaction includes the following steps: (a) Diffusion of aqueous sodium cyanide through the bulk phase and through the resin bulk to active sites (b) Equilibrium exchange of CN- for Br- at the active sites (c) Diffusion of Br- out of the catalyst particle and into the aqueous bulk phase (d) Diffusion of RBr (1-C8H17Br(org)) through the organic bulk phase and through the bulk resin phase to active sites. Some reactions may occur at sites on the catalyst surface, but since the number of surface sites is small compared to the number of sites within the bulk of the catalyst, most of the reaction occurs inside the catalyst bulk. (e) Chemical reaction (intrinsic reaction) between RBr and Resin-PR3+CN- at active sites to produce RCN and Br(f) Diffusion of RCN out of the catalyst particle and into the organic phase. A schematic diagram of the general resin-bound PTC catalyst is given in Figure 13.3.10. As indicated in Figure 13.3.10, spacer chains can increase some reactions by removing the active site away from the polymer chain, and from other active sites. When active sites, particularly quaternary onium salts, are located close to one anFigure 13.3.10. Schematic diagram of the general other, they join to form doublets, triplets, and higher aggregates that are less active resin-bound PTC catalysts. catalyst centers, and that tend to present an “aqueous” face to the reactants. Thus, the use of spacer chains increases the rates of some reactions, such as nucleophilic displacements, two-to-four-fold.1,9,73 When the spacer chain also contains complexable ether oxygen atoms, using 15-crown-5 ether as the PTC functional group, catalyst activity is even greater, as observed in halide exchange of KI with 1-bromooctane.10 Preparation of phase transfer catalyst (PTC) functional groups bound to insoluble resins and their activity for catalyzing two-phase reactions has been extensively stud-

832

Maw-Ling Wang

ied.82,83,97,101 Much work has been done on the preparation and testing of phase-transfer catalysts supported on resins including extensive work by Montanari and co-workers71,74 and by Ford and Tomoi28 and their co-workers. Most published works on resin-bound phase-transfer cataFigure 13.3.11. Crosslink to other polymer chains. lysts use a styrene-divinylbenzene resin (SDV) and related resins, taking advantage of the huge amount of technology available on these resins due to their use as ion-exchange resin supports (Figure 13.3.11). Tomoi and co-workers117 suggest that solvents may affect rates of triphase-catalyzed reaction in three ways: intrinsic chemical reactivity; solvent effect on ion-exchange rate; and overall activity, including diffusion effects due to swelling of polymer-supported phosphonium salts under three-phase conditions. First, the intrinsic activity of the catalysts, as well as of soluble phosphonium salts, depended slightly on organic solvents for cyanide displacement reactions. Second, the exchange rate of chloride ion in the catalysts vs. that of acetate depends on the solvents when the degree of ring substitution is less than 16%. With 30% ring-substituted catalysts, the rate increases and hardly depends on the solvents. Third, the overall catalyst reactivity for the reaction of organic halides with NaCN depends on the substrate and organic solvents. For 1-bromooctane, the catalysts were more reactive in good solvents (e.g., chlorobenzene) than in poor solvent (e.g., octane). Shan and co-workers,98 Wang and Wu136,137 examined the effects of solvents and other resin-bound catalysts parameters (macroporosity, microporosity, crosslink density and size of catalyst pellet). They all show that the swelling in organic solvents is an important factor affecting the conversion of the reactant, as shown in Table 13.3.26.98 Table 13.3.26. Effect of the organic solvent on the yield of ester from benzyl bromide and aqueous KOAc under standardized conditions Catalysta

Solvent (dielectric constant, ε)b

Yield of ester, %

Macro (6%)-400

Cyclohexane (2.02) Toluene (2.38) Chlorobenzene (5.62) Benzyl ethyl ketone (17.4) Nitrobenzene

13.2 16.5 23.3 94.5 100.0

Micro (6%)-400

Cyclohexane (2.02) Toluene (2.38) Chlorobenzene (5.62) Benzyl ethyl ketone (17.4) Nitrobenzene

9.2 13.5 18.2 86.0 100.0

Data obtained from Shan, Kang and Li;98 a6% crosslinking; bPEG-400 used for active sites on catalyst


833

13.3.2.2 Effect of solvents on the reaction in triphase catalysis The disadvantage of using liquid-liquid phase-transfer catalysis (ll-PTC) is in the separation of catalyst from product after reaction. This problem can be overcome using the immobilized catalyst on a solid support (e.g., porous polymer pellet). Simple mechanical separation processes, such as filtration or centrifugation, can be employed to separate the solid catalyst from the product in liquid form. A detailed investigation of the effect of polymer particle on the reaction rate was conducted by Wang and coworkers.131,133-137 Wang and Wu136 studied the reaction of substitution of hexachlorocyclotriphosphazene and 2,2,2-trifluoroethanol in an organic solvent/alkaline solution by triphase catalysis. The polymer, which was prepared from the polymerization of styrene monomer and chloromethylstyrene monomer, served as the support for the immobilization of the catalyst. The effects of the structure of the polymer support, which can be related to the factors of the degree of crosslinking, ring substitution (RS), lipophilicity of the polymer, the chloride density and solvents on the imbibed compositions84,118 on the reaction rate or conversion were investigated. This imbibed composition, influenced by the internal structures of the triphase catalyst particles, affected the reactivities. The interaction of the polymer support pellet and the organic solvents play an important role in determining the reaction rate and the conversion of the reactant. The reaction could be improved to obtain a high reaction rate by using a polar solvent. For investigating the degree of crosslinking of the polymer, the resistance of mass transfer within the catalyst pellet is small. When a smaller degree of crosslinking of the polymer support is used. This is due to the fact that a larger value of the swell of the polymer was obtained when a small degree of crosslinking of the polymer was used. Wang and Yu131 have similar observations for the reaction of allyl bromide and 2,4-dibromophenol under triphase catalysis. A maximum value exists for the degree of swell and the imbibed composition, as shown in Table 13.3.28 for the degree of crosslinking.136 In Table 13.3.27,136 the degree of swell for the polymer support with a 6% crosslinking is larger than that for the two other degrees of crosslinking. This implies that greater amounts of NaOCH2CF3 were imbibed into the catalyst pellet with a 6% crosslinking. The reaction rate is directly related to the amount of the imbibed composition. Also, in Table 13.3.27, the imbibed compositions are affected by the structure of the polymer support. The reactivity of the triphase catalysis can also be determined from the composition imbibed by the particles. It can be observed that the reactivities were highly affected by the lipophilicity of the catalyst pellet for the substitution reaction in the organic phase and the hydrophilicity of the catalyst pellet of the ion exchange in the aqueous phase. For example, the reaction rate in the organic phase was promoted by using a lipophilic polymer support catalyst when the substitution reaction rate was slow. In Table 13.3.27, the amount of chlorobenzene and water imbibed in the macroporous pellet was greater than that in the microporous pellets for most cases.136 However, the macroporous pellet with 10% degree of crosslinking had the least lipophilicity and degree of swelling; therefore, the reactivity was the lowest for the macroporous pellet with a 10% crosslinking among the polymer-support catalysts. The reactivity environments which were created by the lipophilicity and the hydrophilicity of the polymer support plays an important role in determining the reactivity. It is known that the distribution of organic phase and aqueous phase existing in the porous pellet is affected by a change of the ring substitution (RS) of the polymer support.84 Wang and Wu136 prepared three kinds of polymer supports with different numbers of ring

834

Maw-Ling Wang

substitution, such as 10%RS, 20%RS and 49%RS, to analyze the lipophilicity of the polymer support. The order of the lipophilicity was 10%RS > 20%RS> 49%RS, which is the same as the order of swelling. However, a maximum value of the apparent rate constant was obtained for using a 20%RS pellet catalyst among the three kinds of ring substitution polymer pellet. Therefore, it is concluded that the lipophilicity of the polymer cannot be too large to enhance the reaction rate. This is due to the fact that the ion exchange rate is retarded to lower the reaction rate because of using a high lipophilic polymer support. It is concluded that the lipophilicity and the hydrophilicity highly influence the reactivity in triphase catalysis. For a two-phase PTC, it is recognized that the polarity of the organic solvent affects the reaction rate. In general, the reaction rate increases with the augmentation of the polarity of the solvents. Table 13.3.28 shows the effects of the organic solvents on the apparent rate constant, ko,app and ka,app.136 A higher value of the apparent rate constant was obtained using solvent of high polarity. This result is consistent with the swelling and the imbibed compositions that are given in Table 13.3.29.136 Table 13.3.27. Compositions of the imbibed solvents and swelling volume of the triphase catalyst pellet with various polymer structures Triphase catalyst

microporous 2%

microporous 6%

microporous 10%

macroporous 2%

macroporous 6%

macroporous 10%

Conditions

ClC6H5 g

H2O g

ClC6H5 H2O/ClC6H5 2.8M NaOCH2CF3/ClC6H5

1.31 1.23 1.92

0.33 0.67


1.19 1.17 1.90

0.62 0.50


1.06 0.96 1.40


NaOCH2CF3, g (calcd value, g)

Swelling volume ratio

0.40 (0.29)

2.4 2.7 3.6

0.60 (0.22)

2.2 2.8 3.4

0.29 0.50

0.19 (0.22)

2.0 2.1 2.9

1.28 1.33 2.29

0.73 0.50

0.34 (0.22)

3.1 3.8


1.54 1.25 2.2

0.82 0.59

0.42 (0.25)

2.5 3.1 3.8


0.76 1.05 1.28

0.63 0.38

0.17 (0.16)

2.7 2.6

Data obtained from Wang and Wu;136 50 mL of chlorobenzene, 20 mL of water, 0.059 mole of (NPCl2)3, 0.7 meq of catalyst was used, 20oC


835

Table 13.3.28. Effects of the concentrations of NaOCH2CF3 and kind of solvent on the apparent intrinsic rate constants, ko,app and ka,app

Solvent

ko,app for [NaOCH2CF3] (M), (min.meq)

-1

ka,app for [NaOCH2CF3] (M), (min.meq)

1.6 M

2.2 M

2.8 M

1.6 M

2.2 M

2.8 M

CH2Cl2

0.25

0.33

0.58

0.036

0.028

0.036

C6H5Cl

0.063

0.12

0.19

0.017

0.015

0.014

C6H5CH3

0.027

0.056

0.15

0.0055

0.006

0.011

n-C6H14

0.015

0.031

0.092

0.0008

0.0008

0.018

-1

Data obtained from Wang and Wu;136 50 mL of solvent, 20 mL of water, 0.0059 mol of (NPCl2)3, 0.18 meq of macroporous catalyst, 20oC, 40-80 mesh of particle

Table 13.3.29. Effects of solvents on the composition of the imbibed solvents and swelling volume of the triphase catalyst pellet Solvent

Conditions

Solvent g

H2O g

NaOCH2CF3, g (calcd value, g)

Volume ratio

CH2Cl2

CH2Cl2 H2O/CH2Cl2

2.75 2.24

0.96

3.2 3.8

ClC6H5


1.28 1.33 2.29

0.73 0.50

0.34 (0.18)

2.2 3.1 3.8

CH3C6H5 H2O/CH3C6H5 2.8M NaOCH2CF3/CH3C6H5

0.57 0.61 0.37

0.29 0.30

0.01 (0.10)

1.7 2.1 1.8

n-C6H5 H2O/n-C6H5 2.8M NaOCH2CF3/n-C6H5

0 0.10 0.10

0.30 0.15

0.01 (0.05)

1 1.4 1.24

CH3C6H5

n-C6H14

Data obtained from Wang and Wu;136 30 mL of solvent, 0.80 meq of catalyst (1 g), 40-80 mesh of macroporous particle, 20oC

The overall kinetics can be divided into two steps by virtue of the presence of the two practically immiscible liquid phases, i.e., (1) a chemical conversion step in which the active catalyst sites (resin with 2,2,2-trifluoroethanoxide ions) react with hexachlorocyclotriphosphazene in the organic solvent, i.e., yResin+-OCH2CF3(s) + (NPCl2)3(org) → yResin+Cl-(s) + N3P3Cl6-y(OCH2CF3)y(org), y=1-6

[13.3.17]

836

Maw-Ling Wang

(2) the ion exchange step in which the attached catalyst sites are in contact with the aqueous phase, i.e., Resin+Cl-(s) + NaOCH2CF3(aq) → Resin+OCH2CF -3(s) + Na+Cl -(aq)

[13.3.18]

The total moles of the catalyst active sites are S; thus S = [Resin+-OCH2CF3(s) + Resin+Cl -(s)]

[13.3.19]

The reaction rates for (NPCl2)3 in the organic phase and for NaOCH2CF3 in the aqueous phase follow pseudo-first-order kinetics and can be written as − −

d [(NPCl 2 ) 3 ] 0 dt

= k 0, app S[(NPCl 2 ) 3 ] ( org )

d [NaOCH 2 CF3 ]a dt

= k a, app S[NaOCH 2 CF3 ]( org )

[13.3.20]

[13.3.21]

where k0,app and ka,app are the apparent rate constants of (NPCl2)3 per unit amount of catalyst (molar equivalent) in the organic phase for triphase catalysis and the apparent rate constant of NaOCH2CF3 per unit amount of catalyst (molar equivalent) in the aqueous phase for triphase catalysis, respectively. Wang and Yang134 carried out the reaction of 2,4,6-tribromophenol and allyl bromide catalyzed with tributylamine immobilized on the solid styrene-chloromethylstyrene polymer support in an alkaline solution of KOH/chlorobenzene. The experimental results indicate that the swelling power is enhanced in an organic solvent of high polarity. Thus, the reactivity of the reaction is increased with the increase in the polarity of the organic solvents. 13.3.2.3 Effect of volume of organic solvent and water on the reactions in triphase catalysis In investigating the effect of the amount of water, the contents of other components are fixed. Changing the amount of water affects the volume ratio of organic phase to aqueous phase and the concentration of nucleophile in the aqueous phase. For the reaction of hexacyclotriphosphazene and sodium 2,2,2-trifluoroethanoxide catalyzed by tributylamine immobilized on the solid styrene-chloromethylstyrene polymer support catalyst.138-140,157 As shown in Figure 13.3.12,140 the reaction rate is decreased with the increase in the amount of water up to a concentration of sodium 2,2,2-trifluoroethanoxide at 2.8 M. However, the reaction rate is then increased with further increase in the amount of water larger than 2.8 M sodium 2,2,2-trifluoroethanoxide. This result indicates that a high concentration of sodium 2,2,2-trifluoroethanoxide reaction will decrease the reaction rate. The main reason is that the intraparticle diffusion is also affected by the concentration of sodium 2,2,2-trifluoroethanoxide due to changing the amount of water (or the volume ratio of organic phase to aqueous phase).


837

Figure 13.3.12. The apparent rate constants vs. the molar ratios of NaOH/HOCH2CF3; 0.059 mol of (NPCl2)3, 9.6×10-5 mol of TBAB catalyst, 50 mL of chlorobenzene, 20 mL of H2O, 20oC; and (∆) 0.07 mol of HOCH2CF3, (*) 0.058 mol of HOCH2CF3, (O) 0.075 mol of NaOH, (c) 0.063 mol of NaOH (Adapted from Ref. [138], by permission.)

The other experiments, in which the concentrations of the components in the aqueous phase are fixed, were carried out by Wu157. The volume of organic phase is also fixed at 50 mL, in which the amount of catalyst and phosphazene are also fixed. A generalized apparent rate constant (pseudo-first-order rate law) k'i,app is defined as

[

]

k ′i, app = k i, app / Resin (s ) / Va ; i = r, f

[13.3.21]

where Resin(s) and Va indicate the total molar equivalent active sites and the volume of water. The results are given in Table 13.3.30.157 The apparent rate constants k'r,app and k'f,app are increased with the increase in the volume of aqueous phase. These results are explained by low concentration of sodium 2,2,2-trifluoroethanoxide in the organic phase because HOCH2CF3 and NaOCH2CF3 are all insoluble in chlorobenzene. After 4 hours of reaction, only 8% of hexacyclotriphosphasene is reacted. The increase in reaction rate by increasing the volume of aqueous solution is not due to the increase in the concentration of NaOCH2CF3 in the organic phase. However, the mass transfer rate within the particles is obviously affected by increasing the concentration of NaOCH2CF3 in the aqueous phase. Table 13.3.30. Apparent rate constants in various NaOCH2CF3 concentration under constant amount of water or constant amount of NaOCH2CF3 NaOCH2 CF3 (M)

Variation of NaOCH2CF3 NaOCH2CF3 (mole)

kr,appa

Variation of water k'f,app

kf,appa

b

1.3

Volume of water (mL)

kr,appa

kf,appa

k'f,appb

50

0.033

0.0031

8.61

1.6

0.035

0.025

0.005

5.56

40

0.038

0.0033

7.33

2.2

0.0525

0.041

0.005

5.56

30

0.044

0.0041

6.83

2.8

0.07

0.055

20

0.055

0.0050

5.56

a

0.005 -1 b

5.56 -1

-4

Data obtained from Wang and Wu;157 min , Lmin meq×10 ; Reaction conditions: 50 mL of chlorobenzene, 0.0059 moles of (NPCl2)3, 0.175 meq of catalyst, 20oC

838

Maw-Ling Wang

REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52

A. Akelah and D. C. Sherrington, Eur. Polym. J., 18, 301 (1982). B. C. Berris, U. S. Patent 5,030,757 (1991). G. E. Boyd and Q. V. Larson, J. Am. Chem. Soc., 89, 6038 (1967). A. Brandstrom, Adv. Phys. Org. Chem., 15, 267 (1977). L. M. Chen, Y. L. Chen and T. C. Chou, Denki Kagaku, 62(12), 1173 (1994). Y. L. Chen and T. C. Chou, Ind. Eng. Chem. Res., 33(3), 676 (1994). Y. L. Chen and T. C. Chou, J. Appl. Electrochem., 24(5), 434 (1994). T. Chiba and M. Okimoto, J. Org. Chem., 56(21), 6163 (1991). M. S. Chiles, D. D. Jackson and P. C. Reeves, J. Org. Chem., 187, 357 (1980). K. B. Chung and M. Tomoi, J. Polym. Sci., Part A: Polym. Chem., 30(6), 1089 (1992). E. Cordoncillo, E., Monros, G., M. A. Tena, P. Escribano and J. Corda, J. Non-Crystalline Solids, 171(2), 105 (1994). B. P. Czech, M. J. Pugia and R. A. Bartsch, Tetrahedron, 41, 5439 (1985). W. H. Daly, J. D. Caldwell, V. P. Kien and R. Tang, Polym. Prepr., Am. Chem. Soc., Div. Polym. Chem., 23, 145 (1982). O. I. Danilova and S. S. Yufit, Mendeleev Commun., 4, 165 (1993). E. V. Dehmlow and J. Schmidt, Tetrahedron Letters, 2, 95 (1976). E. V. Dehmlow and S. S. Dehmlow, Phase Transfer Catalysis, Verlag Chemie GmbH, Weinheim (1980). E. V. Dehmlow and S. S. Dehmlow, Phase Transfer Catalysis, Weinheim, Deerfield Beach, Florida; Basel, Verlag Chemie, (1983). E. V. Dehmlow and T. Remmler, J. Chem. Res.(S), 72 (1977). E. V. Dehmlow, M. Slopianka and J. Heider, Tetrahedron Letters, 27, 2361 (1977). E. V. Dehmlow, R. Tiesen, Y. Sasson and E. Pross, Tetrahedron, 41(14), 2927 (1985). E. V. Dehmlow, Tetrahedron Letters, 9, 1 (1976). A. K. Dillow, S. L.J. Yun, D. Suleiman, D. L. Boatright, C. L. Liotta, and C. A. Eckert, Ind. Eng. Chem. Res., 35(6), 1801 (1996). J. S. Do and T. C. Chou, J. Appl. Electrochem., 19, 922 (1989). M. Ellwood and J. Griffiths, J. Chem. Soc. Chem. Comm., 181 (1980). T. L. Evans, U. S. Patent 4,520,204 (1985). W. K. Fife and X. Yue, Bull. Chem. Soc. Jpn., 109, 1278 (1987). O. E. Filippova, I. N. Topchieva, V. V. Lutsenko and V. P. Zubov, Vysolomol Soedin., Ser. A, 26, 402 (1984). W. T. Ford and M. Tomoi, Adv. Polym. Sci., 55, 49 (1984). H. H. Freedman, Pure & Appl. Chem., 58(6), 857 (1986). A. Gobbi, D. Landini, A. Maia and S. Petricci, J. Org. Chem., 63 (16), 5356 (1998). G. W. Gokel, ACS Ser., 326, 24 (1987). G. W. Gokel, D. M. Goli and R. A. Schultz, J. Org. Chem., 48, 2837 (1983). H. P. Gregor and J. I. Bergaman, J. Colloid. Sci., 6, 323 (1951). J. M. Harris and M. G. Case, J. Org. Chem., 48, 5390 (1983). J. M. Harris, N. H. Hundley, T. G. Shannon and E. C. Struck, J. Org. Chem., 47, 4789 (1982). A. W. Herriott, and D. Picker, J. Am. Chem. Soc., 97(9), 2345 (1975). T. C. Huang and S. C. Lin, J. Chin. Inst.Chem. Eng., 19, 193 (1988). T. Ido, M. Saiki and S. Goto, Kagaku, Kagaku Ronbunshu, 14, 539 (1988). T. Ido, M. Saiki and S. Goto, Kagaku, Kagaku Ronbunshu, 15, 403 (1989). T. Ido, Y. Kitamura and S. Goto, Kagaku, Kogaku Ronbunshu, 16, 388 (1990). Y. Imai, J. Macromol. Sci., Chem., A15(5), 833 (1981). H. Iwamoto, Tetrahedron Letters, 24, 4703 (1983). H. Iwamoto, H. Kobayashi, P. Murer, T. Sonoda and H. Zollinger, Bull. Chem. Soc. Jpn., 66(9), 2590 (1993). H. Iwamoto, M. Yoshimara and T. Sonodo, Bull. Chem. Soc. Jpn., 56, 791 (1983). D. Jaeger, M. D. Ward and A. K. Dutta, J. Org. Chem., 53, 1577 (1988). Y. S. Jane and J. S. Shih, J. Mol. Catal., 89(1-2), 29 (1994). J. Jarrouse, C. R. Hebd. Seances Acad. Sci., Ser. C232, 1424 (1951). G. Jin, T. Ido and S. Goto, J. Chem. Eng. Jpn., 31(5), 741 (1998). T. Kawai, British Patent 2,219,292 (1989). H. Kobayashi, and T. Sonoda, Chem. Letters, 1185 (1982). Kohjin Co., Ltd., Japanese Kokai Tokkyo Koho JP 58/147,402 (1983). T. N. Konstantinova and V. B. Bojinov, Dyes and Pigments, 39(2), 69 (1998).


839

J. Kuji and T. Okano, J. Syn. Org. Chem. Jpn., 52(4), 276 (1994). C. S. Kuo and J. J. Jwo, J. Org. Chem., 57, 1991 (1992). D. Landini, A. Maia and F. Montanari, J. Am. Chem. Soc., 100, 2796 (1978). M. D. Lang, G. Zhang, X. F. Chen, N. Feng and S. H. Li, J. Appl. Sci., 70(8), 1427 (1998). H. Ledon, Synthesis, 347 (1974). D. G. Lee and V. S. Chang, J. Org. Chem., 43(8), 1532 (1978). D. J. Liaw and P. Chang, J. Appl. Polym. Sci., 63(2), 195 (1997). C. L. Liotta, Application of Macrocyclic Polydentate Ligands to Synthetic Transformations, in Synthetic Multidentate Compounds, R. M. Izatt, J. J. Christensen, eds., Academic Press, New York, 111 (1978). 61 C. L. Liotta, E. M. Burgess, C. C. Ray, E. D. Black and B. E. Fair, ACS Symp. Ser., 326, 15 (1987). 62 C. L. Liotta,, E. M. Burgess and E. D. Black, Polym. Prepr., Am. Chem. Soc., Div. Polym. Chem., 31, 65 (1990). 63 B. L. Liu, A Study of Catalyzed Reaction of Synthesizing of 2-Mercaptobenzimidazole and Its Derivatives", Ph.D. Thesis, Dept. of Chem. Eng., Nat’l Tsing Hua Univ., Hsinchu, Taiwan (1995). 64 E. Maccarone, G. Perrini and M. Torre, Gazzetta Chimica Italiana, 112 (1982). 65 M. Makosza and E. Bialecka, Tetrahedron Letters, 2, 1983 (1977). 66 M. Makosza, Pure Appl. Chem., 43, 439 (1975). 67 M. Makosza, Russ. Chem. Rev., 46, 1151 (1977). 68 C. A. Marques, M. Selva and P. Tundo, J. Chem. Soc., Perkin Trans. I, 529 (1993). 69 K. Mathias and R. A. Vaidya, Bull. Chem. Soc. Jpn., 108, 1093 (1986). 70 E. Mieczynski, A. M. Trzeciak, R. Grzybek and J. J. Ziolkowski, J. Mol. Catal., A: Chemical, 132 (2-3), 203 (1998). 71 H. Molinari, F. Montanari, S. Quici and P. Tundo, J. Am. Chem. Soc., 101, 3920 (1979). 72 E. Monflier, S. Tilloy, G. Fremy, Y. Castanet and A. Mortreux, Tetrahedron Letters, 36(52), 9481 (1995). 73 F. Montanari, S. Quici and P. Tundo, J. Org. Chem., 48, 199 (1983). 74 F. Montanari, D. Landini, A. Maia, S. Quici and P. L. Anelli, ACS Symp. Ser., 326, 54 (1987). 75 P. W. Morgan, Condensation Polymers by Interfacial and Solution Methods, p. 349, Interscience, New York, (1965). 76 R. T. Morrison and R. N. Boyd, Organic Chemistry, p. 702, 5th ed., Allynand Bacon, Inc., Mass, USA (1987). 77 K. Nakamura, S., Nishiyama, S. Tsuruya and M. Masai, J. Mol. Catal., 93(2), 195 (1994). 78 H. Namba, N. Takahashi, K. Abe and M. Saito, Japanese Patent, Jpn. Kokai Tokkyo Koho 63/196547 (1988). 79 R. Neumann and Y. Sasson, J. Org. Chem., 49, 3448 (1984). 80 R. Neumann, S. Dermeik and Y. Sasson, Israel J. Chem., 26, 239 (1985). 81 T. Nishikubo and T. Iziawa, J. Syn. Org. Chem. Jpn., 51(2), 157 (1993). 82 N. Ohtani, Yuki Gosei Kagaku Kyokaishi, 43, 313 (1985). 83 N. Ohtani, Kagaku Kogyo, 39, 331 (1988). 84 N. Ohtani, C. A. Wilkie, A. Nigam and S. C. Regen, Macromolecules, 14, 516 (1981). 85 N. Ohtani, Y. Inoue, A. Nomoto and S. Ohta, React. Polym., 24(1), 73 (1994). 86 N. Ohtani, Y. Inoue, H. Mizuoka and K. Itoh, J. Polymer. Sci., Part A- Polym. Chem., 32(13), 2589 (1994). 87 J. Pielichowski and P. Czub, Angewandte Makromolekulare Chemie, 251, 1 (1997). 88 S. L. Regen, J. Org. Chem., 42, 875 (1977). 89 S. L. Regen, J. Am. Chem. Soc., 97, 5956 (1975). 90 S. L. Regen, J. C. K. Heh and J. McLick, J. Org. Chem., 44, 1961(1979). 91 S. L. Regen, J. J. Bese and J. McLick, J. Am. Chem. Soc., 101, 116 (1979). 92 S. L. Regen and J. J. Besse, J. Am. Chem. Soc., 101, 4059 (1979). 93 G. Rothenberg, L. Feldberg, H. Wiener and Y. Sasson, J. Chem. Soc., Perkin Trans. 2, 11, 2429 (1998). 94 Y. Sasson and R. Neumann, Handbook of Phase Transfer Catalysis, Blackie Academic & Professional, Chapman & Hall, New York (1997). 95 Y. Sasson and S. Zbaida, Ger. Patent DE 3,307,164 (1983). 96 E. G. See, C. A. Buehler and D. E. Pearson, Survey of Organic Synthesis, p. 285, Wiley Interscience, New York (1970). 97 F. Sevc, J. Kahovec and J. Hradil, Chem. Listy, 81, 183 (1987). 98 Y. Shan, R. Kang and W. Li, Ind. Eng. Chem. Res., 28, 1289 (1989). 99 S. S. Shavanon, G. A. Tolstikov and G. A. Viktorov, Zh. Obsch. Khim., 59, 1615 (1989). 100 S. S. Shavanov, G. A. Tolstikov, T. V. Shutenkova, Zh. Obshch. Khim., 57, 1587 (1987). 53 54 55 56 57 58 59 60

840

101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154

Maw-Ling Wang D. C. Sherrington, K. J. Kelly, Org. Coat. Appl. Polym. Sci., Proc., 46, 278 (1981). S. Shimizu, Y. Sasaki and C. Hirai, Bull. Chem. Soc. Jpn., 63, 176 (1990). K. Shinoda, and K. Yasuda, J. Org. Chem., 56(12), 4081 (1991). K. Sjoberg, Aldrichimica Acta, 13, 55 (1980). M. B. Smith, Synth. Commun., 16, 85 (1986). M. Stark, Y. Q. Wang, O. Danielsson, H. Jornvall and J. Johansson, Anal. Biochem., 265(1), 97 (1998). C. M. Starks and C. L. Liotta, Phase Transfer Catalysis, Principles and Techniques, Academic Press, New York (1978). C. M. Starks, J. Am. Chem. Soc., 93, 195 (1971). C. M. Starks, ACS Symposium Series, No. 326, 1 (1987). C. M. Starks, C. L. Liotta and M. Halpern, Phase Transfer Catalysis, Fundamentals, Applications and Indutrial Perspectives, Chapman and Hall, New York (1994). C. M. Starks and R. M. Owens, J.Am. Chem. Soc., 93, 3615 (1973). G. T. Szabo, K. Aranyosi and L. Toke, Acta Chim. Acad. Sci., Hung., 110, 215 (1982). J. Y. Tarng and J. S. Shih, J. Chin. Chem. Soc., 41(1), 81 (1994). L. Toke and G. T. Sazbo, Acta chem. Acad. Sci., Hung., 93, 421 (1977), as summarized by G. E. Totten and N. A. Clinton, Rev. Macromol. Chem. Phys., C28, 293 (1988). L. Toke, G. T. Szabo and K. Somogyi-Werner, Acta Chim. Acad. Sci., Hung., 101, 47 (1979). M. Tomoi and W. T. Ford, J. Am. Chem. Soc., 103, 3821 (1981). M. Tomoi, E. Nakamura, Y. Hosokawa and H. Kakuuichi, J. Polym. Sci., Polym. Chem. Ed., 23, 49 (1985). M. Tomoi, E. Ogawa, Y. Hosokama and H. Kakiuchi, J. Polym. Chem., Polym. Chem., 20, 3421 (1982). A. W. Trochimczuk, A. W. and B. N. Kolarz, Eur. Polym. J., 28(12), 1593 (1992). M. L. Tsai and T. C. Chou, J. Chin. Inst. Chem. Eng., 27(6), 411 (1996). J. Uglestad, T. Ellingsen and A. Beige, Acta Chem. Scand., 20, 1593 (1966). R. Vladea and T. Simandan, Rev. Chim. (Bucharest), 41, 421 (1990). D. H. Wang and H. S. Weng, J. Chin. Inst. Chem. Eng., 27(3), 129 (1996). D. H. Wang and H. S. Weng, J. Chin. Inst. Chem. Eng., 26(3), 147 (1995). D. H. Wang and H. S. Weng, Chem. Eng. Sci., 50(21), 3477 (1995). D. H. Wang and H. S. Weng, Chem. Eng. Sci., 43, 2019 (1988). M. L. Wang and A. H. Liu and J. J. Jwo, Ind. Eng. Chem. Res., 27, 555 (1988). M. L. Wang and B. L. Liu, J.Mol. Catal. A: Chemical, 105, 49 (1996). M. L. Wang, M. L. and B. L. Liu, Int. J. Chem. Kinetics, 28, 885 (1996). M. L. Wang and B. L. Liu, Ind. Eng. Chem. Res., 34(11), 3688 (1995). M. L. Wang, M and C. C. Yu, J. Polymer Sc., Part A: Polym. Chem., 31, 1171 (1993). M. L. Wang and C. C. Yu, J. Chin. Inst. Eng., 23(3), 153 (1992). M. L. Wang, M. L. and C. Z. Peng, J. Appl. Polym. Sci., 52, 701 (1994). M. L. Wang and H. M. Yang, Ind. Eng. Chem. Res., 31(8), 1868 (1992). M. L. Wang and H. M. Yang, J. Mol. Catal., 62, 135 (1990). M. L. Wang and H. S. Wu, Ind. Eng. Chem. Res., 31, 490 (1992). M. L. Wang and H. S. Wu, Ind. Eng. Chem. Res., 31(9), 2238 (1992). M. L. Wang and H. S. Wu, Chem. Eng. Sci., 46, 509 (1991). M. L. Wang and H. S. Wu, J. Chem. Soc., Perkin Trans., 2, 841 (1991). M. L. Wang and H. S. Wu, J. Org. Chem., 55, 2344 (1990). M. L. Wang and K. R. Chang, Ind. Eng. Chem. Res., 29, 40 (1990). M. L. Wang and K. R. Chang, J. Chin. Inst. Eng., 16(5), 675 (1993). M. L. Wang and K. R. Chang, J. Mol. Catal., 67, 147 (1991). M. L. Wang, M. L. and S. W. Chang, Chem. Eng. Commum., 146, 85 (1996). M. L. Wang and S. W. Chang, Bull. Chem. Soc. Jpn., 66, 2149 (1993). M. L. Wang, M. L. and S. W. Chang, Ind. Eng. Chem. Res., 34, 3696 (1995). M. L. Wang and Y. M. Hsieh, Develop. In Chem. Eng. and Min. Proc., 1(4), 225 (1993). M. L. Wang, M. L., C. C. Ou and J. J. Jwo, Ind. Eng. Chem. Res., 33, 2034 (1994). M. L. Wang, M. L., C. C. Ou and J. J. Jwo, J. Mol. Catal., A: Chemical, 99, 153 (1995). M. L. Wang, M. L., C. C. Ou and J. J. Jwo, Bull. Chem. Soc. Jpn., 67, 2249 (1994). Y. M. Wang, Z. P. Zhang, Z. Wang, J. B. Meng and P. Hodge, Chin. J. Polym. Sci., 16(4), 356 (1998). W. P. Weber and G. W. Gokel, Phase Transfer Catalysis in Organic Synthesis, Springer Verlag: Berlin, Heideberg, New York (1977). Y. L. Wen, M. Y. Yeh, Y. S. Lee and Y. P. Shih, J. Chin. Inst. of Chem. Eng., 27(6), 427 (1996). H. S. Weng and D. H. Wang, J. Chin. Inst. of Chem. Eng., 27(6), 419 (1996).

13.4 Effect of polymerization solvent

841

155 C. R. White U. S. Patent 4,642,399 (1987). 156 J. Wild and N. Goetz, Ger. Patent DE 3,820,979 (1989). 157 H. S. Wu, Study on the Displacement Reaction of Phosphazene with Trifluoroethanol by Phase-Transfer Catalysis, Ph.D. Thesis, Department of Chemical Engineering, National Tsing Hua University, Hsinchu, Taiwan (1990). 158 H. S. Wu, Fang, T. R., S. S. Meng and K. H. Hu, J. of Mol. Cata., A: Chemical, 136(2), 135 (1998). 159 M. Yamada, Y. Watabe, T. Sakakibara and R. Sudoh, J. Chem. Soc. Chem. Commun., 179 (1979). 160 B. Yan, Combinatorial Chemistry & High Throughput Screening, 1(4), 215 (1998). 161 S. S. Yufit, Russian Chemical Bulletin, 44(11), 1989 (1995).

13.4 EFFECT OF POLYMERIZATION SOLVENT ON THE CHEMICAL STRUCTURE AND CURING OF AROMATIC POLY(AMIDEIMIDE) Norio Tsubokawa Faculty of Engineering, Niigata University, Niigata, Japan

13.4.1 INTRODUCTION Aromatic poly(amide-imide) (PAI) has an outstanding resistance not only to the thermal operations but also mechanical, electrical, and chemical operations. Although the properties of PAI are inferior to those of aromatic polyimide (PI), which is one of the most heat-resistant polymers, PAI has been widely utilized as a high performance heat resistant polymer as well as PI, because PAI is superior to PI in its workability in industry.1,2 In general, PAI is prepared by the following two processes: diamine process and diisocyanate process. Diisocyanate process is achieved by the direct polycondensation of trimellitic anhydride (TMAH) with aromatic diisocyanate, such as 4,4’-diphenylmethane diisocyanate, in a polar solvent such as N-methyl-2-pyrrolidone (NMP) as shown in Eq. [13.4.1].1-3 On the other hand, the diamine process is achieved by a two-step reaction:1 (1) the polymerization (polycondensation and polyaddition) of trimellitic anhydride chloride (TMAH-CI) with aromatic diamine, such as 4,4’diaminodiphenylether (DDE) or 4,4’diaminodiphenylmethane (DDM) in a polar solvent such as N,N-dimethylacetamide (DMAc), to give poly(amic acid-amide) (PAAA) and (2) the imidation of PAAA by heating as shown in Eq. [13.4.2].

[13.4.1]

842

Norio Tsubokawa

[13.4.2]

PAI is supposed to be a linear polymer containing equivalent amounts of amide and imide bonds. Therefore, PAI is soluble in polar solvents such as NMP, DMAc, and N,N-dimethylformamide (DMF). However, it is well known that PAI becomes insoluble and infusible when it is heated over 200°C. We have pointed out that some remaining carboxyl groups of PAI play an important role in the curing of PAI by heating.4,5 Because of such a thermosetting property, PAI solution is used as a heat resistant temperature coating or an insulating enamel of magnet wire. In the following section, the effect of polymerization solvent, DMAc and mixed solvent of methyl ethyl ketone (MEK) with water,6 on the chemical structure and curing of PAIs prepared by diamine process will be summarized.7 13.4.2 EFFECT OF SOLVENT ON THE CHEMICAL STRUCTURE OF PAI 13.4.2.1 Imide and amide bond content of PAI Four kinds of PAAA, as precursor of PAI, were prepared by the polycondensation of TMAH-Cl with DDE or DDM in DMAc and MEK/water mixed solvent (MEK containing 30 vol% of water) at room temperature. The results are shown in Table 13.4.1. It was found that the conversions reached 78-96% within 2 h at room temperature and the rate of the polycondensation of TMAHCl with diamine in MEK/water mixed solvent was much larger than that in DMAc. Table 13.4.1. Polymerization conditions and conversion of samples [Data from reference 7] Sample No.

Solvent

Diamine

Time, h

Conversion, %

PAAA-1

MEK/H2O

DDE

0.5

90.3

PAAA-2

DMAc

DDE

2.0

89.6

PAAA-3

MEK/H2O

DDM

0.5

78.4

PAAA-4

DMAc

DDM

2.0

96.8

TMAH-Cl=diamine=0.06 mol ; solvent, 172 ml; TEA, 5.0 ml; room temp.

These PAAAs, PAAA-1, PAAA-2, PAAA-3, and PAAA-4, were heated at 180°C for 2 h to give PAI-1, PAI-2, PAI-3, and PAI-4, respectively. Figure 13.4.1 shows the infrared spectra of (A) PAAA-1, (B) PAI-1, and (C) PAI-3.


843

IR spectra of PAAA-1 show absorptions at 1530 and 1660 cm-1, which are characteristic of amide bond, 1510 and 1590 cm-1 which are characteristic of benzene nuclei, and 1230 cm-1, which is characteristic of ether bond, but the absorption at 1380 and 1780 cm-l, which are characteristic of imide bond, are hardly observed. On the other hand, IR spectra of PAI-1 and PAI-3, obtained by heating of PAAA-1 and PAAA-3, respectively, show new absorptions at 1380 and 1780 cm-1. Among the adsorptions of imide bond, the absorption at 1380 cm-1 is assigned to C-N stretching vibrations of all imide bond (cyclic and acyclic imide bond as shown in Eq. [ 13.4.3 ] and 1780 cm-1 is assigned to C=O stretching vibrations of five-member Figure 13.4.1 IR spectra of PAAA-1, PAI-1, and PAI-3 imide rings (cyclic imide bond). The absorption at 1530 cm-1 is assigned to N-H [Data from reference 7] stretching vibrations of amide bond. Furthermore, the absorption at 1510 cm-1 is assigned to benzene nuclei that is stable to heat treatment.

Therefore, the content of cyclic imide, amide and all imide bond of PAI was estimated by the absorbance ratio, D1380/D1510 (the absorbance ratio of absorbance at 1380 cm-1 to that of benzene nuclei), D1780/D1510, and D1530/D1510, respectively. Table 13.4.2 shows the absorbance ratio of cyclic imide, amide, and all imide bond before and after heat treatment of PAAAs. The considerable increase of cyclic and all imide bond content of PAAAs and the decrease of amide bond were observed by heating. But the effect of solvent on the imidation of PAAAs was hardly observed.

844

Norio Tsubokawa

Table 13.4.2. IR absorbance ratio of PAAA and PAI [Data from reference 7] Heat treatment Sample No.

Temperature o C

Time h

PAAA-1 PAI-1

180

2

PAAA-2 PAI-2

180

2

PAAA-3 PAI-3

180

2

PAAA-4 PAI-4

180

2

Absorbance ratio Cyclic imide

Amide

All imide

D1780/D1510

D1530/D1510

D1380/D1510

0.05

0.68

0.37

0.24

0.43

0.57

0.10

0.59

0.36

0.26

0.45

0.58

0.08

0.98

0.66

0.26

0.69

0.82

0.04

0.86

0.29

0.27

0.63

0.79

13.4.2.2 Intrinsic viscosity and carboxyl group content Table 13.4.3 shows the effect of solvent on the intrinsic viscosity, [η], and carboxyl group content of PAIs. The carboxyl content was determined by potentiometric titration. It is interesting to note that intrinsic viscosity of PAI from PAAA prepared in DMAc is larger than that in MEK/water mixed solvent. This suggests that the polymerization degree decreases with decreasing activity of TMAH-Cl in MEK/water mixed solvent because of the hydrolysis of TMAH-Cl by water. Table 13.4.3. Properties of PAI samples [Data from reference 7] Sample No.

Solvent

Diamine

Conversion, %

[η], dl/ga

COOH, eq/gb

PAI-1

MEK/H2O

DDE

90.3

0.36

214

PAI-2

DMAc

DDE

89.6

0.54

44

PAI-3

MEK/H2O

DDM

78.4

0.26

328

PAI-4

DMAc

DDM

96.8

0.32

73

a

Solvent, NMP ; 30.0°C. bDetermined by potentiometric titration

In addition, the content of carboxyl groups in PAIs prepared in MEK/water mixed solvent was considerably larger than that prepared in DMAc. This also suggests the hydrolysis of TMAH-Cl in MEK/water mixed solvent. 13.4.3 EFFECT OF SOLVENT ON THE CURING OF PAI BY HEAT TREATMENT 13.4.3.1 Chemical structure of PAI after heat treatment Table 13.4.4 shows the effect of solvent on the change of chemical structure and formation of insoluble part of PAI-1 and PAI-2 after heat treatment. It became apparent that the con-


845

tent of cyclic imide bond and all imide bonds of PAI-1 and PAI-2 further increased by heating of the corresponding PAAAs at 180°C for 2 h followed by heat treatment at 260°C for 2 h. The content of amide bond decreased by heating of PAAA at 180°C for 2 h, but increased by post-heating at 260°C for 2 h. Table 13.4.4. IR absorbance ratio and insoluble part of samples after heat treatment in air [Data from reference 7] Heat treatment Sample No.

Temperature o C

Time h

PAAA-1

Absorbance ratio

Insoluble part %

Cyclic imide

Amide

All imide

D1780/D1510

D1530/D1510

D1380/D1510

0.05

0.68

0.37

0

PAI-1

180

2

0.24

0.43

0.57

0

PAI-1

260

2

0.33

0.56

0.71

76.7

0.10

0.59

0.36

0

PAAA-2 PAI-2

180

2

0.26

0.45

0.58

0

PAI-2

260

2

0.31

0.50

0.64

16.8

13.4.3.2 Curing PAI by post-heating The formation of the insoluble part in NMP was observed and the amount of insoluble part formed by heating of PAI-1 (obtained in MEK/water mixed solvent) was larger than that by heating of PAI-2 (obtained in DMAc). The increase of imide and amide bond content by post-heating at 260°C is considered as follows: the imidation of amic acid structure may be proceeded by both intermolecular and intramolecular imidation. The latter produces cyclic imide bond, but the former produces acyclic imide bond to give crosslinking material as shown in Eq. [13.4.3]. Since PAIs obtained by heating of PAAAs at 180°C are completely soluble in NMP, the intramolecular imidation preferentially proceeds at 180°C, but intermolecular imidation scarcely proceeds. On the other hand, by post-heating at 260°C, the crosslinking reaction proceeds by the intermolecular imidation of terminal carboxyl groups of PAI with remaining amic-acid structure (Eq. [13.4.4]) and the amide bond in main chain of PAI (Eq. [13.4.5]) to give the insoluble part in NMP. The reaction induced the increase of imide bond content of PAI after post-heating. The increase of amide bond after post-heating may be due to the formation of crosslinking structure by the amidation of terminal amino groups of PAI with carboxyl groups of PAI in main chain (Eq. [13.4.5]-[13.4.8]). Therefore, PAI obtained from the heating of PAAA obtained in MEK/water mixed solvent, which has many carboxyl groups, produces more insoluble part by post-heating. Figure 13.4.2 shows the effect of heating temperature on the curing of PAI obtained by heating of PAAA at 180°C in MEK/water mixed solvent and DMAc. As shown in Figure 13.4.2, by post-heating at 280°C for 2 h in air, insoluble part in NMP reached 100%, indicating the almost complete curing of PAIs. It is interesting to note that PAIs formed by heating of PAAA obtained in MEK/water mixed solvent (PAI-1 and PAI-3) were found to be cured

846

Norio Tsubokawa

more easily than in DMAc (PAI-2 and PAI-4). This is due to the fact that PAI-1 and PAI-3 contains more carboxyl groups than PAI-2 and PAI-4. 13.4.4 CONCLUSIONS The effects of polymerization solvent on the curing of PAIs were investigated and the following results were obtained: (1) The carboxyl group content of PAIs from PAAAs prepared in MEK/H2O mixed solvent was larger than that in Figure 13.4.2. Relationship between heating temperature and insoluble part of PAI samples after heating (2h) in air DMAc. [Data from reference 7]. (2) The curing of PAIs prepared in MEK/H2O proceeded easier than those in DMAc. (3) The curing of PAIs by heating was due to the intermolecular reaction between functional groups of PAIs, such as carboxyl and amino groups. REFERENCES 1 2 3 4 5 6 7

M.Fujita and T.Fujita, Rubber Digest, 18 (1), 21 (1966). S.Terney, J.Keating, J.Zielinski, J.Hakala, and H.Sheffer, J. Polym. Sci., A-1, 8, 683 (1970). Y.Imai and T.Takahasi, Kobunshi Kagaku, 29, 182 (1972). N.Tsubokawa, I.Yamamoto, and Y.Sone, Kobunshi Ronbunshu, 44, 389 (1987). N.Tsubokawa, I.Yamamoto, and Y.Sone, Kobunshi Ronbunshu, 44, 831 (1987). H.Uchiyama and Y.Imai, Kobunshi Kagaku, 28, 73 (1971). N.Tsubokawa, M.Murata, and Y.Sone, Netsu Kokasei Jushi, 12, 12 (1991).

14

Solvent Use in Various Industries Attempts to reduce solvent use in the production of various materials require background information on the current inventory, the reasons for selecting certain solvents, the effect of various solvents on the properties of final products, future trends and the possibilities for solvent replacement. Information on solvent use investigates these areas. This chapter is possible only because of a thorough evaluation by large groups of scientists and engineers assembled by US Environmental Protection Agency. This has produced Compliance Sector Notebooks which contain invaluable information on solvent use. Full documents can be found on the EPA website at http://es.epa.gov/oeca/sector/index.html. These are useful in the investigation of a particular industry. Similar data from other countries were not available but open literature and patents have been included to help the reader to understand changes occurring in other countries.

14.1 ADHESIVES AND SEALANTS George Wypych ChemTec Laboratories, Inc., Toronto, Canada

Adhesives and sealants are manufactured from a variety of polymers. Their selection and their combinations used impact solvent selection. Most solvent systems are designed to optimize the solubility of the primary polymer. Adhesives can be divided into ones which bond by chemical reaction and ones which bond due to physical processes.1 Chemically reactive adhesives are further divided into three more categories for those that bond through polymerization, polyaddition, or polycondensation. Physically bonding adhesives include pressure sensitive and contact adhesives, melt, or solution adhesives, and plastisols. Polymerization adhesives are composed of cyanoacrylates (no solvents), anaerobic adhesives (do not contain solvents but require primers for plastics and some metals which are solutions of copper naphthenate),2 UV-curable adhesives (solvent-free compositions of polyurethanes and epoxy), rubber modified adhesives (variety solvents discussed below). Polyaddition adhesives include epoxy and polyurethane polymers which can either be 100% solids, water-based, reactive or non-reactive hot melts or contain solvents mostly to regulate viscosity. Typical solvents include methyl ethyl ketone, acetone, mineral spirits, toluene, and xylene.3 Polycondensation adhesives include phenol-formaldehyde resin, polyamides, polyesters, silicones and polyimides. With the exception of polyesters (which require ethanol and N-methylpyrrolidone as solvents) and polyimides (which require

848

George Wypych

methyl amyl ketone, butyl acetate, methyl ethyl ketone, 2-ethoxyhexyl acetate as solvents), these adhesives can be made without solvents. Pressure sensitive and contact adhesives are made from a variety of polymers including acrylic acid esters, polyisobutylene, polyesters, polychloroprene, polyurethane, silicone, styrene-butadiene copolymer and natural rubber. With the exception of acrylic acid ester adhesives which can be processed as solutions, emulsions, UV curable 100% solids and silicones (which may contain only traces of solvents), all remaining rubbers are primarily formulated with substantial amounts of solvents such as hydrocarbon solvents (mainly heptane, hexane, naphtha), ketones (mainly acetone and methyl ethyl ketone), and aromatic solvents (mainly toluene and xylene). Melt adhesives and plastisols do not contain solvents. The solution adhesives group includes products made from the following polymer-solvent systems: nitrocellulose (typical solvents include solvent combinations usually of a ketone or an ester, an alcohol and a hydrocarbon selected from isopropanol, 2-butylhexanol, amyl acetate, acetone, methyl ethyl ketone), nitrile rubber (main solvent - methyl ethyl ketone), polychloroprene (which is usually dissolved in a mixture of solvents including a ketone or an ester, an aromatic and aliphatic hydrocarbon selected from naphtha, hexane, acetone, methyl ethyl ketone, benzene, toluene), and polyvinyl acetate (water). In addition to the solvents used in adhesives, solvents are needed for surface preparation4 and primers. Their composition may vary and is usually designed for a particular substrate, often using fast evaporating solvents and environmentally unfriendly materials with significant adverse health effects. Detailed data on the total amount of solvents used by adhesive industry could not be found. The adhesive manufacturing industry continues to grow at a very fast pace. Total adhesive production, according to Frost & Sullivan, was $18.25 billion in 1996 and this is expected to grow to $26.2 billion in 2003.5 Solvent-based materials in 1995 constituted 13% of total production in North America, 14% in Europe, 15% in Japan, and 25% in the Far East.6 Many industries which use solvent-based adhesives have moved to South America and Asia where regulations restricting emissions are less severe.5 The shoe industry is now concentrated in South America. There are many initiatives to decrease solvent emissions. For example, World Bank’s assistance program for developing countries focuses on this issue.5 But in spite of the fact that solvent-based adhesives lost some of their markets (3.3% during the period of 1994-1996),5 they still hold 14-15% of the European market.6 It is estimated that the use of solvents contributes 24% of all VOC emissions. According to one source adhesives were responsible for a 6% share in these emissions in 1993.7 Another source8 blames adhesives for 7% of total VOC emissions in Germany in 1995. Sealants are divided into groups according to the generic names of polymer base. The main groups include: polyurethanes, silicones, acrylics, polysulfide and others (PVC, polybutylene, styrene-butadiene-styrene copolymers, polychloroprene, and several others). The amount of solvent used in sealants is controlled by the standards which previously divided sealants into two groups: these below 10% VOC and those above. Recently, a provision was made to include water-based acrylics and the limit of VOC for class A sealants was increased to 20%. Polyurethane sealants and structural adhesives can be made without solvent (the first solvent-free polyurethane sealant was made in 1994).9 Solvents are added to reduce sealant viscosity and to aid in the manufacture of polymer. Typical solvents used are mineral spirits, toluene, and xylene. A small amount of solvent is emitted from curatives

14.1 Adhesives and sealants

849

which contain methyl ethyl ketone. Most formulations of silicone sealants do not contain solvents. In some sealants, traces of benzene and toluene can be found. Acrylic sealants are water-based but they may also contain ethylene and propylene glycols, mineral spirits and mineral oil. There are also solvent-based acrylic sealants which contain substantial amounts of solvents such as mineral spirits, toluene and xylene. Polysulfide sealants usually contain toluene but methyl ethyl ketone is also used. The group of class B sealants contains substantially more solvents (up to 40% by volume) but there are some exceptions. PVC sealants are based on plastisols and they can be made without solvents. Butyl rubber based sealants usually contain hydrocarbons (C6-C12). Styrene-butadiene-styrene based sealants usually have a large amount of solvents selected from a group including toluene, heptane, hexane, methyl ethyl ketone, isobutyl isobutyrate, n-amyl acetate, n-amyl ketone. They are usually processed in solvent mixtures. Polychloroprene is usually dissolved in a mixture of solvents including ketones or esters, and aromatic and aliphatic hydrocarbons. The list includes naphtha, hexane, acetone, methyl ethyl ketone, benzene, and toluene. The world market of sealants was estimated in 1996 at $2 billion and was expected to grow in 2003 to $2.75 billion with an annual growth rate of 4.5% which is slightly lower than that expected for adhesives (5.3%).5 The changing trends are clearly visible when developments in technology are studied but many barriers to reductions in solvent use exist such as the high investment required, longer processing time, frequently higher material cost of adhesives, and the psychological barrier of changing established adhesive practices. In many instances, adhesive performance is predicted by its superficial characteristics such as strong smell which might suggests that the material has superior properties, its initial green strength which for many indicates good bonding properties, and high viscosity often related to good processing characteristic.10 Since the alternative materials may not have much odor, or require of longer time to reach strength and have a low viscosity, users are suspicious that their potential performance may be inferior. The following information reviews some recent findings which may contribute to future changes. In the shoe industry, a major breakthrough occurred in 1928 when polychloroprene was first introduced.1,10 The first, simple formulation is still manufactured and is used worldwide because the glue can be easily prepared by simply making a solution of the polymer. This gives a product with good adhesion to various substrates. Many new products are available today as potential replacements. Hot melt adhesives can be used in some applications but they still require solvents for cleaning, degreasing, and swelling. Also, their bond strength is frequently inadequate. Reactive systems are not yet used in the shoe industry but reactive hot melts are finding applications. Their broader use is hampered by their sensitivity to moisture which requires special equipment and special care.10 Water-based adhesives are the most likely replacement product. They also need special equipment for processing because of the high heat of evaporation of water (although water based adhesives contain 50% polymer compared with 15-20% in solvent based adhesives).10 Two sport shoe manufacturers, Nike and Reebok, already use this technology. Traditional polychloroprene adhesives can be modified in several ways to be useful in water-based systems. Figure 14.1.1 shows peel strength of several adhesives. The solvent based adhesive (A) has excellent properties both in terms of green strength and bond strength. A simple emulsion of polychloroprene (B) has relatively good ultimate strength

850

George Wypych 60

150 Peel strength, N/25 mm

Peel strength, N/25 mm

1 min 7 day

100

50

0

A

B

C

D

Adhesive type Figure 14.1.1. Green strength and adhesion of several adhesives. Symbols are explained in the text. [Data from B Archer, International J. Adhesion & Adhesives, 18, No.1, 15-8 (1998).]

50 40 30 20 10 0

A

B

C

D

Adhesive type Figure 14.1.2. Contactability of various adhesives. Symbols are explained in the text. [Data from B Archer, International J. Adhesion & Adhesives, 18, No.1, 15-8 (1998).

but lacks green strength and therefore does not meet the performance requirements of shoe manufacturers and other industries.11 Adhesive (C) is a blend of polychloroprene with polyurethane in a water-based system. This modification gives both green strength and peel strength but Figure 14.1.2 shows that peel strength of a freshly applied adhesive is lower than that of solvent based polychloroprene which may cause problems in holding both adhering surfaces together. The adhesive (D) was developed in an interesting new process which involves the emulsification of a solvent based adhesive obtained from styrene-isoprene-styrene rubber.11 After emulsification, the solvent is stripped under vacuum to produce a solvent-free adhesive. The reason for emulsification of the complete adhesive as opposed to emulsification of the rubber alone is to produce a homogeneous adhesive system which would otherwise suffer from separation of polymer particles surrounded by a layer of emulsifying agents. Figures 14.1.1 and 14.1.2 show that this system is superior to solvent-based adhesives. Adhesives can be further improved by polymer blending and by adhesive foaming. A foamed adhesive layer requires less material (approximately 4 times less than a conventional adhesive), it requires less drying time (less water and faster evaporation), and provides an improved bond strength. A foamed adhesive has a larger surface area which increases the surface area of contact with the substrate. The compressed rubber foam has a higher tear strength than unfoamed film of the same thickness.11 Also, regulations are helping to reduce the solvent content of adhesives.12,13 The use of chlorinated solvents, frequently used in primers and for cleaning, has been discontinued based on the Montreal Protocol. From June 1998, the production of a pair of shoes in Europe should not involve the use of more than 20 g of solvent. This is only partially successful solution since shoe production is expected to move out of developed countries to less restrictive jurisdictions. Solvent Emission Directive will continue to restrict solvent use in Europe. Many changes have occurred and more are expected in adhesives based on thermoplastic polyurethanes (TPU).3,14,15In the last 30 years, TPU based adhesives were manufac-

14.1 Adhesives and sealants

851

tured with solvents for shoes, food packaging, and textile and plastic film lamination. Current technologies use TPU in the form of a hot melt, as a reactive PUR and as a thermoplastic laminating film.3 Reactive hot-melts were first introduced in the early 1980s and since than have grown very rapidly. After application the adhesive is cured by moisture.14 These adhesives are already in use by the automotive industry (bonding carpet to door panels, tray assembly, lenses to headlamp housing, and lamination of foam to fabric) and in furniture and building products (moldings, picture frames, decorative foil, edgebanding), in bookbinding, and in the footwear industry. Polyurethane water dispersions are expected to grow 8-10%/year from the current 5,000-6,000 tones/year market in Europe.14 Applications are similar to those of hot melts. UV-curable pressure-sensitive adhesives are the most recent application of the advancing radiation curing technology.16 Low viscosity formulations allow the use of standard application techniques with several advantages such as improved production rate, energy efficiency, improved properties of the final products, and new potential applications for pressure-sensitive additives in thicker films with mechanical performance. It is expected that radiation cured materials will expand at a rate of 10%/year.17 Adhesives constitute 16% by value and 13% by volume of radiation cured products (two major applications for radiation cured materials are coatings and inks). Henkel introduced a series of water-based laminating adhesives.18 Hot melt systems, high-solids solvent systems with a 3 times higher solids content, and water based adhesives have been introduced to textile lamination to replace traditional low-solids solvent-based adhesives.19 Odor elimination is the additional benefit which has helped to drive the replacement of solvent-based systems.20 In packaging materials, most odors are related to the solvents used in inks, coatings and adhesives. Also, coalescing solvents from water-based systems caused odors. Elimination of solvent is a priority but solvent replacement may also change the response to the odor because solvents such as toluene and xylene smell like lubricating oils or turpentine whereas isopropanol smells more like a disinfectant. Odors stem not only from solvents but also from products of the thermal and UV degradation of other components and solvents.20 In view of the above efforts, it is surprising that the majority of recent patents on adhesives are for solvent-based systems.21-26 The new inventions include a universal primer,21 an adhesive composition in which solvents have been selected based on Snyder’s polarity (only solvents which belong to group III are useful in adhesive for automotive applications to avoid a deleterious effect on paint),22 a low VOC adhesive for pipes and fittings,23 a solvent-containing heat-resistant adhesive based on siloxane polyimide,24 a water-based polyimide adhesive,25 and two-component solvent-free polyurethane adhesive system for use in automotive door paneling.26

REFERENCES 1 2 3 4 5 6 7

R Vabrik, G Lepenye, I Tury, I Rusznak, A Vig, International Polym. Sci. Technol., 25, No.3, T/1-9 (1998). D Raftery, M R Smyth, R G Leonard, International J. Adhesion Adhesives, 17, No.4, 349-52 (1997). J B Samms, L Johnson, J. Adhesive Sealant Coun., Spring 1998. Conference proceedings, Adhesive & Sealant Council, Orlando, Fl., 22nd-25th March 1998, p. 87. A Stevenson, D Del Vechio, N Heiburg, Simpson R, Eur. Rubber J., 181, 1, 2829, (1999). K Menzefricke, Adhesive Technol., 15, No.1, 6-7 (1998). L White, Eur. Rubber J., 179, No.4, 24-5 (1997). J Baker, Eur. Chem. News, 69, No.1830, 20-2 (1998).

852

8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

George Wypych P Enenkel, H Bankowsky, M Lokai, K Menzel, W Reich, Pitture e Vernici, 75, No.2, 23-31 (1999). J. van Heumen, H Khalil, W Majewski, G. Nickel, G Wypych, US Patent 5,288,797, Tremco, Ltd., 1994. S Albus, Adhesive Technol., 16, No.1, 30-1 (1999). B Archer, International J. Adhesion & Adhesives, 18, No.1, 15-8 (1998). J C Cardinal, Pitture e Vernici, 74, No.6, 38-9 (1998). Pitture e Vernici, 73, No.20, 39-40 (1997). M Moss, Pigment Resin Technol., 26, No.5, 296-9 (1997). Hughes F, TAPPI 1997 Hot Melt Symposium. Conference Proceedings. TAPPI. Hilton Head, SC, 15th-18th June 1997, p.15-21. D Skinner, Adhesive Technol., 15, No.3, 22-4 (1998). B Gain, Chem. Week, 160, No.14, 28-30 (1998). G Henke, Eur. Adhesives & Sealants, 14, No.1, 18-9 (1997). G Bolte, J. Coated Fabrics, 27, 282-8 (1998). R M Podhajny, Paper, Film & Foil Converter, 72, No.12, 24 (1998). M Levy, US Patent 5,284,510, Paris Laque Service, 1994. I R Owen, US Patent 5,464,888, 3M, 1995. C D Congelio, A M Olah, US Patent 5,859,103, BFGoodrich, 1999. D Zhao, H Sakuyama, T Tomoko, L-C Chang, J-T Lin, US Patent 5,859,181, Nippon Mektron Ltd., 1999. H Ariga, N Futaesaku, H Baba, US Patent 5,663,265, Maruzen Petrochemical Co. Ltd., 1997. E Konig, U F Gronemeier, D Wegener, US Patent 5,672,229, Bayer AG, 1997.

14.2 AEROSPACE George Wypych ChemTec Laboratories, Inc., Toronto, Canada

Figure 14.2.1 shows a schematic diagram of the aerospace manufacturing process.1 Metal finishing is the process in which most solvents and solvent containing materials are used. The main function of the metal finishing process is corrosion protection which requires proper cleaning, surface preparation, and the selection of suitable coatings. The functions of coatings used in aircrafts are different from those used in ordinary coating applications therefore an extrapolation of the progress made with solvent replacement in other coating types is not justified. The typical flight conditions of operating altitude (about 10,000 m above the earth), speed (most frequently 900 km/h), temperature (very low in space at about -60oC and substantially higher after landing up to 80oC), humidity (low in space and high at earth level are combined with condensation due to the tempera-

Figure 14.2.1. Schematic diagram of aerospace manufacturing process. [Reproduced from Profile of the Aerospace Industry. EPA Office of Compliance Sector Notebook Project. US Environmental Protection Agency. November 1998.]

14.2 Aerospace

853

Table 14.2.1. Total releases of solvents by the aerospace industry. [Data from ref. 1] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

62,000

methyl ethyl ketone

995,000

bromotrifluoromethane

750

methyl isobutyl ketone

47,000

n-butyl alcohol

7,000

methyl tetr-butyl ether

550

sec-butyl alcohol

10,000

tetrachloroethylene

285,000

cyclohexane

400

1,1,1-trichloroethane

781,000

1,2-dichlorobenzene

600

trichloroethylene

429,000

1,1-dichloro-1-fluoroethane

10,000

trichlorofluoromethane

1,800

dichloromethane

314,000

toluene

414,000

isopropyl alcohol

1,000

xylene

103,000

methanol

21,000

ture difference), UV radiation (substantially higher during flight), mechanical abrasion due to the high speed of travel, exposure to salt in the atmosphere, exposure to higher level of acids and sulfur dioxide, and exposure to de-icing fluids during winter.2,3 These unusual conditions should be considered in conjunction with the mechanical movement of the coating caused by rapid changes in temperature and the flexing of aircraft elements because of changes in pressure and severe load variations on wings.2 In addition, because of their size, aircrafts must often be painted at low temperatures which requires a coating that will cure at these temperatures without leaving entrapped volatiles. These could evaporate in the low pressure conditions at high altitude and cause the formation of voids where corrosion could start. These factors make the design of an effective coating system a severe technological challenge. Coatings are used by the aerospace industry both for OEM and maintenance purposes. In each case surface cleaning and preparation is required. A paint stripping operation is added to the task in maintenance repainting. Coatings are applied by spraying, brushing, rolling, flow coating, and dipping. Depending on the method of application, the rheological properties of coatings must be adjusted with solvents and, in some cases, with water. An alternative method of viscosity adjustment involves heating the coatings to lower its viscosity by increasing its temperature. This reduces solvent usage. Solvents are also used for equipment cleaning. In addition to paints, sealants are also used. Sealants are mostly based on polysulfides, containing solvents as discussed in the previous section. Also, non-structural adhesives containing solvents are used as gaskets around windows and for carpeting. Paint removal is accomplished by either chemical or blast depainting. Dichloromethane is the most common solvent used for this application. Aerospace industry estimates that 15,000 to 30,000 different materials are used for manufacturing some of which are potentially toxic, volatile, flammable, and contain chlorofluorocarbons. Some of these substances may result in air emissions, waste-waters, and solid waste.

854

George Wypych

Table 14.2.2. Total transfers of solvents by the aerospace industry. [Data from ref. 1] Solvent

Amount, kg/year

Solvent

Amount, kg/year

n-butyl alcohol

2,600

methyl ethyl ketone

500,000

cyclohexane

18,000


17,000

1,2-dichlorobenzene

4,000

tetrachloroethylene

110,000

1,1-dichloro-1-fluoroethane

230


133,000

dichloromethane

68,000

trichloroethylene

98,000


500

trichlorofluoromethane

3,800

ethylene glycol

14,000

toluene

87,000

methanol

12,000

xylene

12,000

Air emissions result from sealing, painting, depainting, bonding, as well as from leakage in storage, mixing, drying, and cleaning. The most common solvents involved in coatings are trichloroethylene, 1,1,1-trichloroethane, toluene, xylene, methyl ethyl ketone, and methyl isobutyl ketone. Wastewater is generated through contamination by paints and solvents used for cleaning operations. Solid waste containing solvents comes from paint overspray intercepted by emission control devices, depainting, cleanup, and disposal of unused paint. Solvents used for cleaning are usually a mixture of dimethyl-benzene, acetone, 4-methyl-2-pentanone, butyl ester of acetic acid, naphtha, ethyl benzene, 2-butanone, toluene and 1-butanol. Some solvents used for painting and cleaning are either recycled are burned to recover energy. In 1996, 199 aerospace facilities (out of 1885 analyzed in the report) released and transferred off-site or discharged to sewers about 12,000 kg of 65 toxic chemicals (solvents in these releases are reported in Table 14.2.1 and transfers in Table 14.2.2). Methyl ethyl ketone, 1,1,1-trichloroethane, trichloroethylene, and toluene accounted for 66% of all releases. 70% of all transfers was for recycling purposes. The aerospace industry released 10,804 tons of VOC in 1997 which constituted 0.61% of the total releases from 29 industries which were analyzed. Thirteen other industries release more VOC than the aerospace industry. Recycling and disposal of solvents in the aerospace industry equals the purchase cost of the solvents. Therefore reduction of solvent use is very cost effective. Some chemical stripping operations are now being replaced by cryogenic stripping with liquid nitrogen. Also, supercritical carbon dioxide has been used in Hughes Aircraft Company in some cleaning applications. Solvent emissions can be reduced through control of evaporation (lids, chillers), by dedicating process equipment (reduces cleaning frequency), production scheduling, immediate cleaning of equipment, better operating procedures, reuse of solvent waste, and use of optimized equipment for paint application. There are plans to evaluate powder coatings and water-based paints.1,2 There are trials to use water as the paint thinner and to lower viscosity of paints by application of resins which have lower viscosity.2 Work is under the way to replace dichloromethane/phenol stripper with benzyl alcohol.2 The introduction of an intermediate layer between the primer and the top coat has been proposed.

14.3 Asphalt compounding

855

This will aid the stripping action of the proposed stripping solvent, benzyl alcohol. VOC have already been reduced in several components: bonding primer (from 1030 to 850 g/l), undercoats (from 670 to 350), top coats (from 700-900 to 250-800), clear coats (from 700-800 to 250-520), surface cleaners (from 850 to 250) as well as other materials.3

REFERENCES 1 2 3

Profile of the Aerospace Industry. EPA Office of Compliance Sector Notebook Project. US Environmental Protection Agency. November 1998. R W Blackford, Surface Coatings International, 80, No.12, 564-7 (1997). R Blackford, Polym. Paint Colour J., 186, No.4377, 22-4 (1996).

14.3 ASPHALT COMPOUNDING George Wypych ChemTec Laboratories, Inc., Toronto, Canada

Numerous construction products are formulated from asphalt and coal tar for such applications as driveway sealers, cutback asphalts, flashing cements, concrete primers, concrete cold mixes, roof cements, expansion joint fillers, patch liquids, waterproofing liquid-applied membranes, and pipeline coatings. All these products are likely to contain solvents. The simplest formulations are mixtures of asphalt and (usually) mineral spirits used for sealing , priming, and coating of concrete. These are usually very low performance products which are used in large quantities because of their low price. They release about 40% of their weight to atmosphere during and after application. Since they do not perform well they have to be re-applied at frequent intervals. Driveway sealer is an example of a product which is used every spring, in spite of the fact that, in addition to the pollution it causes, it also produces a gradual degradation and cracking of the driveway. The only solution for elimination of this unnecessary pollution seems to be banning the product by regulation. Some of these products can be replaced by asphalt emulsions which contain water in place of organic solvents. Several products are used for patching and joint filling purposes. These materials (flashing cement, roof cement, patch liquid, and expansion joint filler) also use solvents to regulate viscosity. The solvents are usually mineral spirits, fuel oil, or polycyclic aromatic hydrocarbons. In addition to the base components, inexpensive fillers such as calcium carbonate or limestone but also still asbestos are added. These products harden on evaporation of the solvent and fill the joints, adhere to surfaces, and provide some waterproofing. These are again, low technology materials, traditionally used because of their very low cost. Most of these products can be replaced by modern sealants which will result in higher initial cost but longer service. The most technologically advanced products are used for waterproofing and pipeline coatings. These products are also based on dispersion of asphalt in the above mentioned solvents but reinforced with addition of polymer. The addition of polymer modifies the plastic behavior of asphalt and renders it elastomeric. Additional solvents are usually added to improve the solubility of polymeric components. Reactive polyurethanes are the most frequently used modifiers for waterproofing liquid membranes. Toluene and xylene are the

856

M Matsumoto, S Isken, J A M de Bont

most frequently used additional solvents. These materials partially solidify because of evaporation of the solvent. Their elastomeric properties are derived from chain extension and crosslinking reactions which form an internal polymeric network which reinforces asphalt. There is no data available on the solvent emissions from these materials but their scale of production suggests that their emissions are probably comparable with the entire rubber industry. This is one industry which should be closely monitored not only because of the emission of the above listed solvents but because some of the low grade solvents used contain large quantities of benzene and hexane. It is also cause for concern that asphalt and tar have carcinogenic components. Recent inventions1-5 are driven by product improvement needs and environmental aspects of application of these products. Janoski’s patent1 describes a product which is an anhydrous blend of polymer and asphalt and is substantially solvent-free. This technology shows that it is possible for an ingenious designer to produce low viscosity materials without using solvents but by selecting the appropriate type and concentration of bituminous materials, polyurethane components, and plasticizers. In another invention,2 a modifier is introduced to increase the adhesion of asphalt/water emulsions to aggregates. Emulsified asphalt is not so deleterious to the environment but its performance suffers from aggregate delamination. In yet another recent invention,3 terpene solvent, which is a naturally occurring (but never in this high concentration), biodegradable material, was used to replace the mineral spirits, xylene, trichloroethane, toluene, or methyl ethyl ketone normally used in cutback formulations (cutback asphalt is a dispersion of asphalt in a suitable solvent to reduce viscosity and allow for cold application). The two other patents4,5 discuss inventions leading to an improvement of high and low temperature properties of asphalt with no special impact on reduction of solvents used.


R J Janoski, US Patent 5,319,008, Tremco, Inc., 1994. P Schilling, E Crews, US Patent 5,772,749, Westaco Corporation, 1998. R W Paradise, US Patent 5,362,316, Imperbel America Corporation, 1994. M P Doyle, J L Stevens, US Patent 5,496,400, Vinzoyl Petroleum Co., 1996. M P Doyle, US Patent 5,749,953, Vinzoyl Technical Services, LLC, 1998.

14.4 BIOTECHNOLOGY 14.4.1 ORGANIC SOLVENTS IN MICROBIAL PRODUCTION PROCESSES

Michiaki Matsumoto, Sonja Isken, Jan A. M. de Bont Division of Industrial Microbiology Department of Food Technology and Nutritional Sciences Wageningen University, Wageningen, The Netherlands

14.4.1.1 Introduction Solvents are not dominating compounds in the biosphere of our planet. Under natural conditions, their presence in appreciable amounts is restricted to specific areas. Only a very limited number of solvents is of biological origin and some may reach higher concentrations in nature. The best known example is ethanol. However, also butanol and acetone can be

14.4.1 Organic solvents in microbial production

857

formed readily by microbes and locally high concentrations may occur. In fact, in the beginning of the 20th century, very large production facilities were in operation for the microbial production of butanol and acetone. Furthermore, terpenes are natural solvents that are produced mainly by plants and locally they can reach high concentrations. For instance, limonene is present in tiny droplets in the peel of oranges. All these solvents are toxic to microbial cells. Some others as higher hydrocarbons that are present for instance in olive oil, are not toxic to microbes as will be discussed later. With the advent of the chemical industry, this picture has changed dramatically. In polluted locations, microorganisms may be confronted with a large number of solvents at high concentrations. With a few exceptions only, it has turned out that microbes can be found that are able to degrade these compounds if their concentration is low. This degradative potential is not unexpected in view of trace amounts that may be present locally in the natural biosphere. But the exposure of cells to unnatural high concentrations of these solvents usually leads to irreversible inactivation and finally to their death. The chemical industry is largely based on solvent-based processes. But in biotechnological processes, the microbes usually are exploited in a water-based system. This approach is quite understandable in view of the preference of microbes for water and the problems solvents pose to whole cells. Solvents often are used to extract products from the aqueous phase but only after the production process has been completed. At this stage, damage to whole cells is obviously no longer relevant. In both, chemical industry and biotechnology, organic solvents have many advantages over water because of the nature of either product or substrate. Consequently, during the last decades many possibilities have been investigated to use solvents in biocatalytic processes.1,2 The more simple the biocatalytic system, the less complex it is to use solvents. Free or immobilized enzymes have been exploited already in a number of systems. Here, biocatalysis may take place in reversed micelles or in an aqueous phase in contact with an organic solvent.3 In a powdered state some enzymes are able to function in pure organic solvents.4 Furthermore, modified enzymes such as polymer bound enzymes5 or surfactant-coated enzymes6 have been developed so that they can solubilize in organic solvents to overcome diffusion limitation. The advantages of enzymatic reactions using organic solvents can be briefly summarized as follows:1,3,4 1) hydrophobic substances can be used; 2) synthetic reactions can take place; 3) substrate or production inhibition can be diminished and 4) bioproducts and biocatalysts can easily be recovered from the systems containing organic solvents. Although organic solvents have often been used in enzymatic reactions, the application of organic solvents for biotransformation with whole-cell systems is still limited. Cells might be continuously in direct contact with the organic phase in a two-phase water-solvent system during the whole production cycle (Figure 14.4.1.1).7 Alternatively, cells may remain separated from the bulk organic phase by using membrane bioreactors (Figure14.4.1.2).8 In these instances, cells encounter phase toxicity9 or molecular toxicity, respectively. Because whole bacterial cells are more complex than enzymes, they pose by far greater problems in operating bioproduction processes when organic solvents are present. The most critical problem is the inherent toxicity of solvents to living organisms.1,2,10,11 As

858


Fig.14.4.1.1. Schematic diagram of a two-phase bioreactor system for continuous 1-octanol production. [After reference 7]. 1-Octanol is produced from n-octane in hexadecene by Pseudomonas oleovorans or recombinant strains containing the alkane oxidation genes.

Fig.14.4.1.2. Schematic diagram of a two-phase hollow-fiber membrane bioreactor system for hydrolytic epoxide resolution. [After reference 8]. The yeast cells contain an epoxide hydrolase that enantioselectively hydrolyzes racemic epoxide resulting in enantiopure epoxide that partitions to the organic phase. Diol produced partitions to the water phase.


859

some solutions have been already found in this area, further progress is expected in the near future. A key problem is the selection of useful solvents in combination with a suitable microbe. Many solvents may be considered with their specific process properties, and many microorganisms may be considered also with their specific advantages or disadvantages. Only on the basis of a detailed understanding of the functioning of a microbial cell in the presence of solvents will it be possible to make a rational selection for a good combination solvent/organism. In the following section we will discuss these key questions. We will also describe recently found solvent-tolerant bacteria and will provide some examples of biotransformation using solvent-tolerant bacteria. 14.4.1.2 Toxicity of organic solvents The toxicity of an organic solvent is closely related to its hydrophobicity, as expressed by the log PO/W values,12-16 the logarithm of the partition coefficient of the solvent between 1-octanol and water (Table 14.4.1). In general, the Gram-negative bacteria show relatively higher solvent tolerances than the Gram-positive bacteria.12,14 This may be caused by the difference in the composition of the cell envelope. Solvents with a high log PO/W resulted in the highest activities, but the range of log PO/W tolerated by an organism is dependent on the type of microorganism. In the following section we will describe what happens in the cell in the presence of solvents.

+

+

+

decane

6.0

+

+

+

+

+

+

+

+

-

nonane

5.5

+

+

+

+

+

+

+

+

-

n-hexyl ether

5.1

+

+

+

+

+

+

+

+

-

octane

4.9

+

+

+

+

+

+

+

+

-

isooctane

4.8

+

+

+

+

+

+

+

-

-

cyclooctane

4.5

+

+

+

+

+

+

-

-

-

diphenyl ether

4.2

+

+

+

+

+

-

-

-

-

n-hexane

3.9

+

+

+

+

+

-

-

-

-

propylbenzene

3.7

+

+

+

+

-

-

-

-

-

o-dichlorobenzene

3.5

+

+

+

-

-

-

-

-

-

cyclohexane

3.4

+

+

+

-

-

-

-

-

-

ethylbenzene

3.2

+

+

-

-

-

-

-

-

-

ATCC26602

+

Saccharomyces uvarum

+

AHU1219

+

Bacillus subtilis

+

Agrobacterium tumefaciens IFO3058

Achromobacter delicatulus LAM1433

+

JCM1474

Escherichia coli IFO3806

+

Alcaligenes faecalis

Pseudomonas fluorescens IFO3507

7.0

IFO3738

Pseudomonas putida IH-2000

dodecane

Pseudomonas putida

Solvent

log PO/W

Table 14.4.1. Relationship between the growth of a cell exposed to an organic solvent and the value of log PO/W of the solvent. [After reference 14]

-

-

-

styrene

2.9

+

-

-

-

-

-

-

-

-

toluene

2.6

+

-

-

-

-

-

-

-

-

benzene

2.1

-

-

-

-

-

-

-

-

-

ATCC26602

-

AHU1219

-

Bacillus subtilis

-

Agrobacterium tumefaciens IFO3058

-

JCM1474

+

Alcaligenes faecalis

+

Achromobacter delicatulus LAM1433

Pseudomonas putida

3.1

Pseudomonas fluorescens IFO3507

Pseudomonas putida IH-2000

p-xylene

IFO3738

Solvent

log PO/W

Saccharomyces uvarum


Escherichia coli IFO3806

860

+, growth; -, no growth

Bar9 suggested that the toxicity in two-phase systems was caused by both the presence of a second phase (phase toxicity) and solvent molecules which dissolved in the aqueous phase (molecular toxicity). Basically, both mechanisms are governed by the same principle in that the solvent accumulates in the microbial membrane. In case of the direct contact between cells and pure solvent, the rate of entry of solvents in a membrane will be very high. If the solvent has to diffuse via the water phase, then the accumulation in membranes will be slower. This latter mechanism on the molecular toxicity has been investigated in more detail.17 In experiments with liposomes from E. coli, and ten representative organic solvents labeled by 14C under aqueous-saturating levels, it was observed that the solvents accumulate preferentially in the cell membrane. The partition coefficients (log PM/B) of the solvents between the model liposome membrane and buffer correlate with those (log PO/W) in a standard 1-octanol-water system: log PM/B = 0.97 x log PO/W - 0.64

[14.4.1.1]17

The accumulation of an organic solvent in the membrane causes changes in the membrane structure. Organic solvents residing in the hydrophobic part of the membrane disturb the interactions between the acyl chains of the phospholipids. This leads to a modification of membrane fluidity which eventually results in the swelling of the bilayer.10 In addition to this, conformations of the membrane-embedded proteins may be altered.10 These changes in the integrity of the membrane also affect the membrane function. The principal functions of the cytoplasmic membrane involve: 1) barrier function, 2) energy transduction and 3) formation of a matrix for proteins. The disruption of lipid-lipid and lipid-protein interactions by the accumulation of organic solvents has a strong effect on the membrane’s function as a selective barrier for ions and hydrophilic molecules. Permeability is of particular importance for protons because the leakage of protons directly affects the primary energy transducing properties of the membrane. The initial rates of proton influx in the absence and presence of different amounts of hydrocarbon were measured.17 The permeability for protons increases with increasing amounts of hydrocarbon. Hence, leakage of protons occurs in the presence of organic sol-


861

Figure 14.4.1.3 Schematic presentation of possible mechanisms of solvent tolerance. A Changes in the structure of cytoplasmic membrane. B Changes in the structure of outer membrane. C Transformation of the solvent. D Active export of the solvents [After reference 32]

vents.17 Not only the impairment of the barrier function is caused by the alterations that occur in the membrane structure when it interacts with organic solvents. It is well known that the activities of the proteins embedded in the membrane are regulated by the membrane thickness, head group hydration, fluidity and fatty acid composition.18,19 All these parameters are also known to be affected by the accumulation of organic solvents. The effects of solvents on these parameters were reviewed by Sikkema et al.10

862


14.4.1.3 Solvent-tolerant bacteria As described in the previous section, the organic solvents with 1 < log PO/W < 4 are considered to be toxic to microorganisms. In 1989, Inoue and Horikoshi21 found a toluene-tolerant Pseudomonas putida strain that grew in a two-phase toluene-water system (log PO/W = 2.5 for toluene). This finding was surprising and went against the dominant paradigm at that time. Solvent tolerance was confirmed by other strains of P. putida22-26 and by other representatives of the genus Pseudomonas.27-30 Furthermore, solvent tolerance has been found in the strains of Gram-positive bacteria Bacillus31,32 and Rhodococcus.33 The key question now is: How do solvent-tolerant bacteria overcome the toxic effects of organic solvents? Some of the possible mechanisms involved in solvent tolerance according to various researchers are shown in Fig.14.4.1.3.32 Current research on changes in the structure of the cytoplasmic membrane shows the involvement of: 1) the composition of the fatty acids of the phospholipids like the cis/trans isomerization of unsaturated fatty acids; 2) composition of phospholipid headgroups and 3) rate of turnover of membrane components. Organic solvents cause a shift in the ratio of saturated to unsaturated fatty acids.34,35 In a solvent-tolerant strain, an increase in the saturation degree has been observed during adaptation to the presence of toluene. Solvent-tolerant strains also have the ability to synthesize trans-unsaturated fatty acids from the cis-form in response to the presence of organic solvents.34,36-38 Increases in the saturation degree and the ratio of trans-form change the fluidity of the membrane and the swelling effects caused by solvents are depressed. Alterations in the headgroups of lipids during the adaptation to solvents have also been observed in some solvent-tolerant strains.37,39 The changes in the composition of the headgroups cause changes in the affinity of the lipids with the organic solvents and in the stability of membrane due to an alteration of bilayer surface charge density. These changes compensate the effect caused by the solvents. In one strain, the rate of phospholipid synthesis increases after exposure to a solvent.40 This strain has a repairing system which is faster than the rate of damage caused by the organic solvent. Unlike Gram-positive bacteria, Gram-negative bacteria such as Pseudomonas have an outer membrane. The outer membrane has been shown to play a role in the protection of the cell from solvent toxicity. Ions such as Mg2+ or Ca2+ stabilize the organization of the outer membrane and contribute to solvent tolerance.38 Low cell surface hydrophobicity caused by changes in the lipopolysaccharide (LPS) content has been reported to serve as a defensive mechanism.41,42 It has also been reported that the porins which are embedded in the outer membrane are relevant to solvent tolerance.37,42-44 The metabolism of organic solvents in solvent-tolerant strains contributes to solvent tolerance by degradation of the toxic compounds. This contribution, however, is considered to be limited33,45 because many solvent-tolerant strains show non-specific tolerance against various organic compounds. Non-specific tolerance to toxic compounds is well known in the field of antibiotic resistance. A wide range of structurally dissimilar antibiotics can be exported out of the cell by multidrug-efflux pumps. Could the export of organic solvents contribute to solvent tolerance? Isken and de Bont46 conducted experiments to determine whether the solvent tolerant Pseudomonas putida S12 was able to export toluene by monitoring the accumulation of 14C labeled toluene in the cells. Toluene-adapted cells were able to export toluene from their


863

Figure 14.4.1.4. Schematic picture of toluene penetration and efflux in the solvent-tolerant Pseudomonas putida S12 [After reference 20].

membranes whereas the non-adapted cells were not. Furthermore, it was observed that in the presence of energy coupling inhibitors, toluene accumulation was the same as in the non-adapted cells. The amount of toluene in the cell was concluded to be kept at a low level by an active efflux system. The presence of a toluene-efflux system is supported by genetic research.47,48 The pump has a striking resemblance to multidrug-efflux systems. Active efflux pumps for solvents have also been detected in other Pseudomonas strains.25,26,30,37 It is obvious that solvent tolerance is caused by a combination of the mechanisms described above. Figure 14.4.1.4 shows a schematic picture of toluene penetration and efflux in the solvent tolerant strain P. putida S12.20 Toluene enters the cell through the outer membrane. At present, it is unclear whether toluene passes through porins or through the phospholipid part of the cell. The efflux pump recognizes and interacts with toluene in the cytoplasmic membrane. Toluene is then pumped into the extracellular medium. 14.4.1.4 Biotransformation using solvent-tolerant microorganisms Many important fine chemicals, including catechols, phenols, aldehydes and ketones, low molecular epoxides and diepoxides, medium-chain alcohols, and terpenoids fall within the range of 1 < log PO/W < 4. The discovery of solvent-tolerant bacteria leads to the new possibility of biocatalytic reaction systems containing organic solvents. By using solvent-tolerant bacteria, a variety of fine chemicals can be formed in microbial production processes. The organic solvents used so far in published research had to be very hydrophobic (logPO/W > 5) in order to prevent microbial inactivation.1 Consequently, many fine chemicals can not beneficially produced in the presence of such solvents because they simply would remain in the water phase and would not partition to the organic phase. The use of solvent-tolerant microorganisms enables the use of less hydrophobic solvents (2.5 < logPO/W < 4) and in such a system, chemicals with a 1 < logPO/W < 4 preferentially will go into the organic phase. Up to now, however, only a few applications using solvent-tolerant microorganisms have been reported. Aono et al.49 reported the oxidative bioconversion of cholesterol as a model biocatalytic reaction using a solvent-tolerant Pseudomonas species. Cholesterol and its

864


products are insoluble in an aqueous solution but dissolve in some organic solvents. The attempt was successful. The conversion of cholesterol was more than 98% and the yield of oxidative products was 80%. Speelmans et al.50 reported on the bioconversion of limonene to perillic acid by a solvent-tolerant Pseudomonas putida. The microbial toxicity of limonene is known to be very high. It is a major component of citrus essential oil and is a cheap and readily available base material. By using a solvent-tolerant strain perillic acid was obtained at a high concentration. This finding brings commercial production nearer. The applications of solvent-tolerant strains in microbial production processes are at present limited, but two strategic options are currently available to use such bacteria.20 Relevant genes can be introduced into solvent-tolerant organisms in order to produce the required product. This approach has been followed successfully by J. Wery in our laboratory who employed an 1-octanol-aqueous system. Methylcatechol was produced from toluene by solvent tolerant P. putida S12. Alternatively, the efflux pump can be expressed in a suitable solvent-sensitive host which would then be more tolerant for a particular solvent. Other benefits may arrive from solvent-resistant bacteria. Ogino et al.26 isolated Pseudomonas aeruginosa LST-03 which can grow in organic solvents with logPO/W >2.4 and secrets organic solvent-stable lipolytic enzymes. They were able to purify an organic solvent-stable protease which was more stable than the commercially available proteases.51 Hence, solvent-tolerant strains have become a source for new enzymes.52 In the near future, the use of solvent-tolerant strains will make the application of organic solvents in biotransformations by whole cells a more realistic option. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

R. León, P. Fernandes, H.M. Pinheiro, and J.M.S. Cabral, Enz. Microb. Technol., 23, 483 (1998). B. Angelova and H.S. Schmauder, J. Biotechnol., 67, 13 (1999). C. Laane, S. Boeren, K. Vos, and C. Veeger, Biotechnol. Bioeng., 30, 81 (1987). A.M. Klibanov, Trends Biotechnol., 15, 97 (1997). Y. Inada, A. Matsushima, M. Hiroto, H. Nishimura, and Y. Kodera, Methods Enzymol., 242, 65 (1994). Y. Okahata, Y. Fujimoto, and K. Iijiro, J. Org. Chem., 60, 2244 (1995). R.G. Mathys, A. Schmid, and B. Witholt, Biotechnol. Bioeng., 64, 459 (1999). W.Y. Choi, C.Y. Choi, J.A.M. de Bont, and C.A.G.M. Weijers, Appl. Microbiol. Biotechnol., 53, 7 (1999). R. Bar, J. Chem. Technol. Biotechnol., 43, 49 (1988). J. Sikkema, J.A.M. de Bont, and B. Poolman, Microbiol. Rev., 59, 201 (1995). M.D. Lilly and J.M. Woodley, in Biocatalysis in organic synthesis, J. Tramper, H.C. van der Plas, and P. Linko, Ed., Elsevier, Amsterdam, 1985, pp.179-192. M. Vermuë, J. Sikkema, A. Verheul, and J. Tramper, Biotechnol. Bioeng., 42, 747 (1993). S.D. Doig, A.T. Boam, D.J. Leak, A.G. Livingston, and D.C. Stuckey, Biocatal. Biotransform., 16, 27 (1998). A. Inoue and K. Horikoshi, J. Ferment. Bioeng., 71, 194 (1991). A.J. Harrop, M.D. Hocknull, and M.D. Lilly, Biotechnol. Lett., 11, 807 (1989). A.N. Rajagopal, Enz. Microb. Technol., 19, 606 (1996). J. Sikkema, J.A.M. de Bont, and B. Poolman, J. Biol. Chem., 269, 8022 (1994). H. Sandermann, Jr., Biochim. Biophys. Acta, 515, 209 (1978). P.L. Yeagle, FASEB J., 3, 1833 (1989). J.A.M. de Bont, Trends Biotechnol., 16, 493 (1998). A. Inoue and K. Horikoshi, Nature, 338, 264 (1989). D.L. Cruden, J.H. Wolfram, R.D. Rogers, and D.T. Gibson, Appl. Environ. Microbiol., 58, 2723 (1992). F.J. Weber, L.P. Ooykaas, R.M.W. Schemen, S. Hartmans, and J.A.M. de Bont, Appl. Environ. Microbiol., 59, 3502 (1993). J.L. Ramos, E. Deque, M.J. Huertas, and A. Haïdour, J. Bacteriol., 177, 3911 (1995). K. Kim, S.J. Lee, K.H. Lee, and D.B. Lim, J. Bacteriol., 180, 3692 (1998).

14.4.2 Solvent-resistant microorganisms

26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52

865

F. Fukumori, H. Hirayama, H. Takami, A. Inoue, and K. Horikoshi, Extremophiles, 2, 395 (1998). H. Nakajima, H. Kobayashi, R. Aono, and K. Horikoshi, Biosci. Biotechnol. Biochem., 56, 1872 (1992). H. Ogino, K. Miyamoto, and H. Ishikawa, Appl. Environ. Microbiol., 60, 3884 (1994). Y. Yoshida, Y. Ikura, and T. Kudo, Biosci. Biotechnol. Biochem., 61, 46 (1997). X.Z. Li, L. Zhang, and K. Poole, J. Bacteriol., 180, 2987 (1998). K. Moriya and K. Horikoshi, J. Ferment. Bioeng., 76, 397 (1993). S. Isken and J.A.M. de Bont, Extremophiles, 2, 229 (1998). M.L. Paje, B.A. Neilan, and I. Couperwhite, Microbiology, 143, 2975 (1997). F.J. Weber, S. Isken, and J.A.M. de Bont, Microbiology, 140, 2013 (1994). H.C. Pinkart, J.W. Wolfram, R. Rogers, and D.C. White, Appl. Environ. Microbiol., 62, 1129 (1996). H.J. Heipieper, G. Meulenbeld, Q. van Oirschot, and J.A.M. de Bont, Chemosphere, 30, 1041 (1995). J.L. Ramos, E. Duque, J.J. Rodriguez-Herva, P. Godoy, A. Haidour, F. Reyes, and A. Fernandez-Barrero, J. Biol. Chem., 272, 3887 (1997). F.J. Weber and J.A.M. de Bont, Biochim. Biophys. Acta, 1286, 225 (1996). V. Pedrotta and B. Witholt, J. Bacteriol., 181, 3256 (1999). H.C. Pinkart and D.C. White, J. Bacteriol., 179, 4219 (1997). R. Aono and H. Kobayashi, Appl. Environ. Microbiol., 63, 3637 (1997). H. Kobayashi, H. Takami, H. Hirayama, K. Kobata, R. Usami, and K. Horikoshi, J. Bacteriol., 181, 4493 (1999). L. Li, T. Komatsu, A. Inoue, and K. Horikoshi, Biosci. Biotechnol. Biochem., 59, 2358 (1995). H. Asano, K. Kobayashi, and R. Aono, Appl. Environ. Microbiol., 65, 294 (1999). G. Mosqueda, M.-S. Ramos-González, and J.L. Ramos, Gene, 232, 69 (1999). S. Isken and J.A.M. de Bont, J. Bacteriol., 178, 6056 (1996). J. Kieboom, J.J. Dennis, G.J. Zylstra, and J.A.M. de Bont, J. Biol. Chem., 273, 85 (1998). J. Kieboom, J.J. Dennis, G.J. Zylstra, and J.A.M. de Bont, J. Bacteriol., 180, 6769 (1998). R. Aono, N. Doukyu, H. Kobayashi, H. Nakajima, and K. Horikoshi, Appl. Environ. Microbiol., 60, 2518 (1994). G. Speelmans, A. Bijlsma, and G. Eggink, Appl. Microbiol. Biotechnol., 50, 538 (1998). H. Ogino, F. Watanabe, M. Yamada, S. Nakagawa, T. Hirose, A. Noguchi, M. Yasuda, and H. Ishikawa, J. Biosci. Bioeng., 87, 61 (1999). N. Doukyu and R. Aono, Appl. Environ. Microbiol., 64, 1929 (1998).

14.4.2 SOLVENT-RESISTANT MICROORGANISMS

Tilman Hahn, Konrad Botzenhart Institut fòr Allgemeine Hygiene und Umwelthygiene Universit¬t Tòbingen, Tòbingen, Germany

14.4.2.1 Introduction Several main properties of microorganisms in relation to solvents can be considered: • Toxic or antimicrobial effects of solvents • Solvent resistance or adaptation of microorganisms • Metabolic activities of microorganisms Antimicrobial effects or solvent-resistant microorganisms are the main topic of this section. 14.4.2.2 Toxicity of solvents for microorganisms 14.4.2.2.1 Spectrum of microorganisms and solvents The growth-inhibiting effects of several solvents on microorganisms are described.1,2 Organic solvents have toxic effect on microorganisms. Table 14.4.2.1 summarizes the relevant organic solvents and their toxicity concerning selected microorganisms.

866

Tilman Hahn, Konrad Botzenhart

Table 14.4.2.1. Toxicity of organic solvents - examples Solvent

Microorganisms

References

Toluene, benzene, ethylbenzene, propylbenzene, xylene, hexane, cyclohexane

Pseudomonas putida

Isken et al. (1999)3 Gibson et al. (1970)4

Terpenes, e.g., alpha-pinene, limonene, β-pinene, terpinolene

Bacillus sp., Saccharomyces cerevisiae, isolated mitochondria

Andrews et al. (1980)5 Uribe et al. (1985)6

Styrene

soil microorganisms

Hartmans et al. (1990)7

Cyclohexane

yeast cells, isolated mitochondria

Uribe et al. (1990)8

Aromatic hydrocarbons

isolated bacterial and liposomal membranes

Sikkema et al. (1992)9 Sikkema et al. (1994)1

Ethanol

yeasts

Cartwright et al. (1986)10 Leao and van Uden (1984)11

14.4.2.2.2 Mechanisms of solvent toxicity for microorganisms The toxicity of organic solvents or hydrophobic substances for microorganisms depends mainly on their effects on biological membranes1,9,12-14 - similar to membrane effects of several anesthetics. This concerns especially effects on cytoplasmatic membranes. The following main changes of membrane structures and functions have been observed: • Accumulation of hydrophobic substances such as organic solvents in cytoplasmatic membranes. This accumulation causes structural and functional changes in the cytoplasmatic membranes and microbial cells. • Structural changes in cytoplasmatic membranes, e.g., swelling of membrane bilayers, increase of surface and thickness of the membranes, changes in the composition of the membrane (e.g., changes in the fatty acid composition), modification of the microviscosity, damage of membrane structures (see below). • Loss of membrane integrity, especially disruption of cytoplasmatic membranes, less damage of outer membranes. Because of these damages often complex cellular structures (e.g., vesicula) or cell functions (decrease of respiratory activities of mitochondria) are destroyed or inhibited. • Interactions of the accumulated lipophilic substances with the cytoplasmatic membranes and especially hydrophobic parts of the cell or cell membranes. Lipid-lipid interactions and interactions between proteins and lipids of the membrane structure (lipid bilayers, membrane-embedded proteins) are discussed. • Effects on passive and active membrane transport systems, e.g., increase of passive efflux and flux of ions such as protons, cations Mg++ and Ca++ or small molecules, stimulation of the leakage of protons and potassium, changes in the uptake of compounds (e.g., solvents) and excretion (e.g., metabolic products), inhibition of active transport systems (e.g., ATP depletion). • Damage of cellular homeostasis and cell physiology, e.g., reduction of transmembrane electrical potentials and proton chemical potentials or proton motive forces as a result of membrane changes (efflux of ions), changes of pH gradients.


867

• Changes in enzyme activities, e.g., inhibition of oxidases and depletion of ATP. Of special relevance are various interactions with enzymes (proteins) in the membrane, e.g., lipid-protein interactions. • Loss of particular cellular functions, e.g., respiratory system of mitochondria or active transport systems (see above). • Loss of complex cell functions, e.g., reduced growth rates and activities of microorganisms. The extent of solvent toxicity to microorganisms is determined by various factors: (a) Hydrophobic or lipophilic properties of solvents. The toxicity of solvents to microorganisms can be described by a partition coefficient (log Po/w) between organic compounds (solvents) and water which is specific for the applied substance. This partition coefficient is based on a standard octanol-water system model.1,15,16 The toxicity and the affinity of solvents to cell structures increase with hydrophobic properties of solvents, e.g., high toxicities with Po/w values of 1-5.1,14 The partition coefficient correlates with the membrane-buffer partition coefficient between membrane and aqueous system.1,14,16 They also depend on membrane characteristics.17,18 (b) Accumulation, partitioning and concentrations of solvents in cell structures (membranes). Dissolution and partitioning of solvents depend essentially on solvent properties, e.g., polarity (specific partition coefficients), or membrane characteristics (influence on partition coefficients). Both dissolution and partitioning can be influenced by additional factors, e.g., cosolvents. The effects on microorganisms can depend typically on solvent concentrations, e.g., dose response effects. (c) Biomass, ratio of concentrations of solvents and biomass. Effect of solvents depends on this ratio. (d) Surrounding conditions, e.g., temperature which influences the proton leakage and microbial activities.19 The toxicity of solvents for microorganisms shows positive and negative consequences, e.g.: • Positive aspects such as antibacterial effects20 which are found in several products. • Negative aspects such as reduced stability of biotransformation and bioremediation processes because of the inactivation of microorganisms. 14.4.2.3 Adaption of microorganisms to solvents - solvent-resistant microorganisms 14.4.2.3.1 Spectrum of solvent-resistant microorganisms Different microorganisms are able to adapt and even to grow in the presence of solvents. Some relevant examples are given in Table 14.4.2.2. Table 14.4.2.2. Solvent-tolerant microorganisms and their resistance to organic solvents Solvent-tolerant microorganisms bacteria from deep sea (1.168 m, Japan)

Solvents benzene, toluene, p-xylene, biphenyl, naphthalene

References Abe et al. (1995)21

868


Solvent-tolerant microorganisms

Solvents

References

deep sea isolates, Flavobacterium strain DS-711, Bacillus strain DS-994

hydrocarbons

Moriya and Horikoshi (1993)22

marine yeasts

n-alkanes

Fukamaki et al. (1994)23

Pseudomonas putida

toluene, m-, p-xylene, 1,2,4-trimethylbenzene, 3-ethyltoluene

Inoue and Horikoshi (1989)24 Cruden et al. (1992)25

Pseudomonas sp.

alpha-pinene

Sikkema et al. (1995)14

E. coli K-12

p-xylene

Aono et al. (1991)26

The concentration of organic solvents which can be tolerated by the microorganisms varies extensively, e.g., growth of Pseudomonas putida in the presence of more than 50 % toluene27 but tolerance of E. coli K-12 in the presence of only up to 10 % p-xylene.26 The growth characteristics of different species can differ to a great extent, e.g., no growth, growth with or without metabolizing solvents. Generally higher solvent tolerance of Gram-negative bacteria compared to Gram-positive bacteria is observed.28 Some solvent-tolerant microorganisms cannot use organic solvents as substrate for growth and need other substrates in complex media.26,27 Other microorganisms can use organic solvents in minimal media as a source of energy or carbon, e.g., Pseudomonas putida in the presence of xylene or toluene.25 14.4.2.3.2 Adaption mechanisms of microorganisms to solvents The mechanisms of solvent tolerance are only partly known.29-31 Relevant microbial adaptation mechanisms are: - Changes in the composition of the cytoplasmatic membrane. The compounds of the membranes such as lipids or proteins can influence the membrane characteristics and, therefore, the adaption to solvents. Mainly phospholipids in the membrane bilayer determine the partitioning of solutes and especially the resistance to solvents17,18 A reduction of the partition coefficient has been observed when the fatty acid composition was changed.32 An increase of monounsaturated fatty acids and a decrease of saturated fatty acids correlated with a higher ethanol tolerance of S. cerevisiae, E. coli, and Lactobacillus strains.33-35 An increase of unsaturated fatty acids is induced by polar solvents and low temperatures, an increase of saturated fatty acids is connected with more apolar solvents and high temperatures.36 Even changes in the configuration of fatty acids, which are provoked by solvents, can lead to adaption mechanisms, e.g., cis-trans conversions.29 - Changes in the microbial structure. Various structural changes can cause a reduction of toxic solvent effects, e.g., the increase of membrane fluidity which is connected with an increase of unsaturated fatty acids (see above) and results in a decrease of the membrane permeability. - Specific structural characteristics of microorganisms. Typical structures of microorganisms vary according to the microbial species, e.g., outer membrane characteristics of bacteria. For instance, Gram-negative bacteria such as Pseudomonas sp. tolerate higher concentrations of hydrophobic compounds compared to Gram-positive bacteria. The resistance of the outer membrane correlates with the solvent-tolerance.28


869

- Alterations of the cell envelope structure (cell wall). Mechanical alterations and chemical modifications of the cell wall can reduce the microbial resistance to solvents. The most interesting chemical modification concerns hydrophobic or hydrophilic abilities of the cell wall.37 Decreasing hydrophobicity of the cell wall enhances the adaption of microorganisms to solvents.38 - Suppression of the effects of solvents on membrane stability. - Limitation of solvent diffusion into the cell (see above). - Repairing mechanisms, e.g. enhanced phospholipid biosynthesis. - Transport or export systems. The excretion of compounds out of the microbial cell and cytoplasmatic membrane is well known but only documented for some substances, e.g., for drugs.39 Passive and active transport systems are relevant, e.g., ATP driven systems. Export systems for the several solvents must be assumed. - Immobilization and mobilization of microorganisms and solvents. The adsorption of solvents to microorganisms can be reduced if the contact is decreased. For instance, immobilization of microorganisms or solvents minimizes the contact. An immobilization and reduction of toxicity was shown if adsorption materials were added.40 - Surrounding conditions, e.g., low temperature, which can induce higher solvent resistance of microorganisms (see above). 14.4.2.4 Solvents and microorganisms in the environment and industry examples Microorganisms are frequently observed in organic-aqueous systems containing solvents are essential in natural and in industrial processes. The occurrence and role of microorganisms and organic compounds in these two-phase organic-aqueous systems are similar to the effects described above (see Section 14.4.2.2). Although toxic effects on microorganisms in these natural and industrial processes are well known, reliable data concerning solvent-resistant microorganisms are not available. 14.4.2.4.1 Examples 14.4.2.4.1.1 Biofilms, biofouling, biocorrosion Important examples for organic-aqueous systems are surface-associated biofilms which are a form of existence of microorganisms. Microorganisms, mostly bacteria, are embedded into a glycocalyx matrix of these biofilms.41,42 This biofilm matrix mediates the adhesion of microorganisms to surfaces, concentrates substances and protects microorganisms from antimicrobial agents.41,42 Several organic-aqueous systems can be observed, especially surface of surrounding materials (pipes, etc.) in relation to water or ingredients (e.g., oil in pipes) related to water between ingredients and surrounding materials (pipes, etc.). Some aspects of solvents in these organic-aqueous biofilm systems are studied. Solvents can occur in water systems emitted from surrounding organic materials.43 It was shown that solvents are important concerning microbial biocorrosion and biofouling processes, e.g., by swelling and hydrolysis of materials.44 Despite these well-known aspects, reliable data and studies concerning solvent-resistant microorganisms in biofilm, biofouling, or biocorrossion processes are not shown. Nevertheless similar mechanisms in biofilms must be assumed as described above (see Section 14.4.2.3) because similar conditions occur (organic-aqueous systems).

870


14.4.2.4.1.2 Antimicrobial effects, microbial test systems The toxic mechanisms of solvents to microorganisms described above (see Section 14.4.2.2) are frequently used in effects of antimicrobial agents. The damage of microbial biomembranes is fundamentally connected with the antimicrobial effects of several solvents on bacteria.45 Biomembranes and other microbial structures can be affected by solvents via similar processes as described. Relevant examples are naked viruses which are generally more resistant to viruzi agents because the envelopes of viruses are damaged by viruzi substances such as some solvents. Another example of solvent-like interactions are effects of antimicrobial agents to capsules of bacterial spores, e.g., Bacillus species. Bacterial or enzymatic toxicity tests are used to assay the activity of organic compounds including solvents. A survey of environmental bacterial or enzymatic test systems is given by Bitton and Koopman.46 The principles of these test systems are based on bacterial properties (growth, viability, bioluminescence, etc.) or enzymatic activities and biosynthesis. The toxicity of several solvents were tested in bacterial or enzymatic systems, e.g., pure solvents such as phenol in growth inhibition assays (Aeromonas sp.),46 solvents in complex compounds such as oil derivates,46,47 solvents in environmental samples such as sediments or solvents used in the test systems.46,48,49 The efficiency of several test systems, e.g., Microtox tests or ATP assays, vary, e.g., looking at the effects of solvents.46 14.4.2.4.1.3 Industrial processes In industrial processes the main microbial activities connected with solvents are: • Processes in biotechnology, biotransformation and biocatalysis,50,51 e.g., production of chemicals from hydrophobic substrates or use of solvents as starting materials for microbiological reactions. • Bioremediation: degradation of environmental pollutants, e.g., wastewater treatment or bioremediation in biofilm reactors.52-55 Various microorganisms and microbial mechanisms are relevant in these industrial processes. Examples are conversion processes of organic substances, e.g., by bacterial oxygenases.56,57 Normally low-molecular-weight aromatic hydrocarbons including solvents are converted in these biotransformation processes.58 References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

J. Sikkema, J.A.M. deBont, B.Poolman, J. Biol. Chem., 269, 8022 (1994). G.J. Salter, D.B. Kell, Crit. Rev. Biotechnol., 15, 139 (1995). S. Isken, A. Derks, P.F.G. Wolffs, J.A.M. deBont, Appl. Environ. Microbiol., 65, 2631 (1999). D.T. Gibson, G.E. Cardini, F.C. Maseles, R.E. Kallio, Biochemistry, 9, 1631 (1970). R.E. Andrews, L.W. Parks, K.D. Spence, Appl. Environ. Microbiol., 40, 301 (1980). S. Uribe, J. Ramirez, A. Pena, J. Bacteriol., 161, 1195 (1985). S. Hartmans, M.J. van der Werf, J.A.M. de Bont, Appl. Environ. Microbiol., 56, 1347 (1990). S. Uribe, P. Rangel, G. Espinola, G. Aguirre, Appl. Environ. Microbiol., 56, 2114 (1990). J. Sikkema, B. Poolman, W.N. Konigs, J.A.M. deBont, J. Bacteriol., 174, 2986 (1992). C.P. Cartwright, J.R. Juroszek, M.J. Beavan, F.M.S. Ruby, S.M.F. DeMorais, A.H. Rose, J. gen. Microbiol., 132, 369 (1986). C. Leao, N. vanUden, Biochim. Biophys. Acta, 774, 43 (1984). M.J. DeSmet, J. Kingma, B. Witholt, Biochem. Biophys. Acta, 506, 64 (1978). M.R. Smith in Biochemistryof Microbial Degradation, C. Ratledge, Ed, Kluwer Academic Press, Dordrecht, 1993, pp. 347-378. J. Sikkema, J.A.M. deBont, B. Poolman, B., FEMS Microbiol. Rev., 59, 201 (1995). A. Leo, C. Hansch, D. Elkins, Chem. Rev., 71, 525 (1971).


16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58

871

W.R. Lieb and W.D. Stein in Transport and Diffusion Across Cell Membranes, W.D. Stein, Ed., Academic Press, N.Y., 1986, pp. 69-112. M.C. Antunes-Madeira, V.M.C. Madeira, Biochim. Biophys. Acta, 861, 159 (1986). M.C. Antunes-Madeira, V.M.C. Madeira, Biochim. Biophys. Acta, 901, 61 (1987). W. DeVrij, A. Bulthuis, W.N. Konings, J. Bacteriol., 170, 2359 (1988). F.M. Harold, Adv. Microb. Physiol., 4, 45 (1970). A. Abe, A. Inoue, R. Usami, K. Moriya, K. Horikoshi, Biosci. Biotechnol. Biochem., 59, 1154 (1995). K. Moriya, K. Horikoshi, J. Ferment. Bioeng., 76, 168 (1993). T. Fukamaki, A. Inoue, K. Moriya, K. Horikoshi, Biosci. Biotechnol. Biochem., 58, 1784 (1994). A. Inoue, K. Horikoshi, Nature, 338, 264 (1989). D.L. Cruden, J.H. Wolfram, R. Rogers, D.T. Gibson, Appl. Environ. Microbiol., 58, 2723 (1992). R. Aono, K. Albe, A. Inoue, K. Horikoshi, Agric. Biol. Chem., 55, 1935 (1991). A. Inoue, M. Yamamoto, K. Horikoshi, K., Appl. Environ. Microbiol., 57, 1560 (1991). A.J. Harrop, M.D. Hocknull, M.D. Lilly, Biotechnol. Lett., 11, 807 (1989). H.J. Heipieper, R. Diefenbach, H. Keweloh, Appl. Environ. Microbiol., 58, 1847 (1992). H.C. Pinkart, D.C. White, J. Bacteriol., 179, 4219 (1997). J.L. Ramos, E. Duque, J. Rodriguez-Herva, P. Godoy, A. Haidour, F. Reyes, A. Fernandez-Barrero, J. Biol. Chem., 272, 3887 (1997). M.C. Antunes-Madeira, V.M.C., Madeira, Biochim. Biophys. Acta, 778, 49 (1984). P. Mishra, S. Kaur, Appl. Microbiol. Biotechnol., 34, 697 (1991). L.O. Ingram, J. Bacteriol., 125, 670 (1976). K. Uchida, Agric. Biol. Chem., 39, 1515 (1975). L.O. Ingram, Appl. Environ. Microbiol., 33, 1233 (1977). V. Jarlier, H. Nikaido, FEMS Microbiol. Lett., 123, 11 (1994). Y.S. Park, H.N. Chang, B.H. Kim, Biotechnol. Lett., 10, 261 (1988). D. Molenaar, T. Bolhuis, T. Abee, B. Poolman, W.N. Konings, J. Bacteriol., 174, 3118 (1992). H. Bettmann, H.J. Rehm, Appl. Microbiol. Biotechnol., 20, 285 (1984). J.W. Costerton, J. Ind. Microbiol. Biotechnol., 22, 551 (1999). J.W. Costerton, Int. J. Antimicrobial Agents, 11, 217 (1999). K. Frensch, H.F. Scholer, D. Schoenen, Zbl. Hyg. Umweltmed., 190, 72 (1990). W. Sand, Int. Biodeterioration Biodegradation, 40, 183 (1997). K. Wallhäußer, Praxis der Sterilisation, Desinfektion, Konservierung, Thieme Verlag, Stuttgart, 1998. G. Bitton, B. Koopman, Rev. Environ. Contamin. Toxicol., 125, 1 (1992). K.L.E. Kaiser, J.M. Ribo, Tox. Assess, 3, 195 (1988). B.J. Dutka, K.K. Kwan, Tox. Assess, 3, 303 (1988). M.H. Schiewe, E.G. Hawk, D.I. Actor, M.M. Krahn, Can. J. Fish Aquat., 42, 1244 (1985). O. Favre-Bulle, J. Schouten, J. Kingma, B. Witholt, Biotechnology, 9, 367, 1991. C. Laane, R. Boeren, R. Hilhorst, C. Veeger, in C. Laane, J. Tramper, M.D. Lilly, Eds., Biocatalysis in organic media, Elsevier Publishers, B.V., Amsterdam, 1987. C. Adami, R. Kummel, Gefahrstoffe Reinhaltung Luft, 57, 365 (1997). A.R. Pedersen, S. Moller, S. Molin, E. Arvin, Biotechnol. Bioengn., 54, 131 (1997). P.J. Hirl, R.L. Irvine, Wat. Sci. Technol., 35, 49 (1997). M.W. Fitch, D. Weissman, P. Phelps, G. Georgiou, Wat. Res., 30, 2655 (1996). S.V. Ley, F. Sternfeld, S.C. Taylor, Tetrahedron Lett., 28, 225 (1987). G.M. Whited, W.R. McComble, L.D. Kwart, D.T. Gibson, J. Bacteriol., 166, 1028 (1986). M.R. Smith, J. Bacteriol., 170, 2891 (1990)

872

Tsuneo Yamane

14.4.3 CHOICE OF SOLVENT FOR ENZYMATIC REACTION IN ORGANIC SOLVENT

Tsuneo Yamane Graduate School of Bio- and Agro-Sciences, Nagoya University, Nagoya, Japan

14.4.3.1 Introduction The ability of enzymes to catalyze useful synthetic biotransformations in organic media is now beyond doubt. There are some advantages in using enzymes in organic media as opposed to aqueous medium, including 1) shifting thermodynamic equilibrium to favor synthesis over hydrolysis, 2) reduction in water-dependent side reaction, 3) immobilization of the enzyme is often unnecessary (even if it is desired, merely physical deposition onto solid surfaces is enough), 4) elimination of microbial contamination, 5) suitable for reaction of substrates insoluble and/or unstable in water, etc. Here organic media as the reaction system are classified into two categories: substrates dissolved in neat organic solvents and solvent-free liquid substrates. Although the latter seems preferable to the former, if it works, there are a number of cases where the former is the system of choice: for example, when the substrate is solid at the temperature of the reaction, when high concentration of the substrate is inhibitory for the reaction, when the solvent used gives better environment (accelerating effect) for the enzyme, and so forth. Prior to carrying out an enzymatic reaction in an organic solvent, one faces choice of a suitable solvent in the vast kinds of organic solvents. From active basic researches having been carried out in the past two decades, there has been a remarkable progress in our understanding of properties of enzymes in organic media, and in how organic solvents affect them. Some researchers call the achievement ‘medium engineering’. A comprehensive monograph was published in 1996 reviewing the progress.1 In this article organic solvents often used for enzymatic reactions are roughly classified, followed by influence of solvent properties on enzymatic reactions and then properties of the enzymes affected by the nature of the organic solvents are briefly summarized. 14.4.3.2 Classification of organic solvents Among numerous kinds of organic solvents, the ones often used for enzymatic reactions are not so many, and may be classified into three categories (Table 14.4.3.1),2 in view of the importance of water content of the organic solvent concerned (see Section 14.4.3.3). 1) water-miscible organic solvents These organic solvents are miscible with water at the temperature of the reaction. Any cosolvent system having 0 - 100% ratio of the solvent/water can be prepared from this kind of solvent. Note that some organic solvents having limited water solubility at ambient temperature, and hence are not regarded as water-miscible, become miscible at elevated temperature. 2) Water-immiscible organic solvents

14.4.3 Choice of solvent for enzymatic reaction

873

Table 14.4.3.1. Classification of solvents commonly used for enzymatic reactions in organic media. [Adapted, by permission, from T. Yamane, Nippon Nogeikagaku Kaishi, 65, 1104(1991)] 1) Water-miscible organic solvents Methanol, ethanol, ethylene glycol, glycerol, N,N’-dimethylformamide, dimethylsulfoxide, acetone, formaldehyde, dioxane, etc. 2) Water-immiscible organic solvents (water solubility [g/l] at the temperature indicated)

•

alcohols (n-, iso-) propyl alcohol, (n-, s-, t-) butyl alcohol, (n-, s-, t-) amyl alcohol, n-octanol, etc. • esters methyl acetate, ethyl acetate (37.8, 40oC), n-butyl acetate, hexyl acetate, etc. • alkyl halides dichloromethane (2, 30oC), chloroform, carbon tetrachloride, trichloroethane (0.4, 40oC), etc. • ethers diethyl ether (12, 20oC), dipropyl ether, diisopropyl ether, dibutyl ether, dipentyl ether, etc. 3) Water-insoluble organic solvents (water solubility [ppm] at the temperature indicated)

• • •

aliphatic hydrocarbons n-hexane (320, 40oC), n-heptane (310, 30oC), isooctane (180, 30oC), etc. aromatic hydrocarbons benzene (1200, 40oC), toluene (880, 30oC), etc. allicyclic hydrocarbons cyclohexane (160, 30oC), etc.

These organic solvents have noticeable but limited solubility of water, ranging roughly 0.1 - 10 % its solubility. The water solubility is of course increased as the temperature is raised. 3) Water-insoluble organic solvents These solvents are also water-immiscible and have very low water solubility so that they are regarded as water-insoluble, i.e., water is practically insoluble in the organic solvents. Most aliphatic and aromatic hydrocarbons belong to this category. In Table 14.4.3.1 organic solvents often used for enzymatic reactions are listed together with their water solubilities (although not for all of them). 14.4.3.3 Influence of solvent parameters on nature of enzymatic reactions in organic media Factors that influence the activity and stability of enzymes in organic media have been mostly elucidated. Several of them are mentioned below. 1) Water activity, aw Among factors, the amount of water existing in the reaction system is no doubt the most influential. To emphasize the effect of water, the author once proposed to say ‘enzymatic reaction in microaqueous organic solvent’, instead of merely say ‘enzymatic reaction in organic solvent’.3-5

874

Tsuneo Yamane

Very trace amount of water or nearly anhydrous state renders practically no enzymatic reaction. In this context, it should be reminded that commercially available enzyme preparations, or the enzymes even after lyophilization or other drying procedures, contain some water bound to the enzyme proteins. Whereas, excess water in the reaction system results in hydrolysis of the substrate, which is often unfavorable side reaction, giving rise to lower yield of product. Thus, there exist usually the optimal water content for each enzymatic reaction of concern. Water molecules in the microaqueous system exist in three different states: 1) water bound to the enzyme protein, 2) water dissolved freely in the solvent (plus dissolved substrate), and 3) water bound to impurities existing in the enzyme preparation or bound to the support materials if immobilized enzyme particles are used. Therefore, the following equation with respect to water holds: Total water = (Water bound to the enzyme) + (Water dissolved in the solvent)

[14.4.3.1]

+ (Water bound to the immobilization support or to impurities of the enzyme preparation)

Water affecting most of the catalytic activity of the enzyme is the one bound to the enzyme protein.6 From the above equation, it can be well understood that the effect of water varies depending on the amount of enzyme used and/or its purity, kind of solvent, and nature of immobilization support, etc. as far as the total water content is used as the sole variable. Also, it is often asked what is the minimal water content sufficient for enzymatic activity? It should be recognized that a relation between the degree of hydration of the enzyme and its catalytic activity changes continuously. There exist a thermodynamic isotherm-type equilibrium between the protein-bound water and freely dissolved water, and its relationship is quite different between water-miscible and water-insoluble solvents.4 A parameter better than the water content, water activity, aw, was proposed to generalize the degree of hydration of a biocatalyst in organic media.7 aw is a thermodynamic parameter which determines how much water is bound to the enzyme and in turn decides the catalytic activity to a large extent among different kinds of the organic solvents. aw is especially useful when water-insoluble organic solvent is used because the precise water content is hard to be measured due to its low solubility. It was shown that profiles between aw Figure 14.4.3.1 Activity of Lipozyme catalyst as a function of water activity in a range of solvents. [Adapted, by and the reaction rate were similar when the permission, from R.H. Valivety, P.J. Halling and A.R. same reaction was carried out in different Macrae, Biochim. Biophys. Acta, 1118, 221 (1992)]. solvents at varying water contents.8 In dif-


875

ferent solvents, maximum reaction rate was observed at widely different water content, but if water content was qualified in terms of aw, the optimum was observed at almost the same aw (Figure 14.4.3.1). However, as seen from Figure 14.4.3.1, the profile does not lay on a single curve, and the absolute optimal reaction rate varies depending on the kind of the solvent, implying that aw is not almighty. 2) Hydrophobicity (or polarity), logP A hydrophobicity parameter, logP, was first proposed for microbial epoxidation of propene and 1-butene.9,10 logP is the logarithm of P, where P is defined as the partition coefficient of a given compound in the standard n-octanol/water two phase: P=

solubility of a given compound in n − octanol phase solubility of a given compound in water phase

[14.4.3.2]

Laane et al.10 concluded as a general rule that biocatalysis in organic solvents is low in polar solvents having a logP < 2, is moderate in solvents having a logP between 2 and 4, and is high in apolar solvents having a logP > 4. They also stated that this correlation between polarity and activity paralleled the ability of organic solvents to distort the essential water layer bound to the enzyme that stabilized the enzyme. Since logP can easily be determined experimentally, or be estimated from hydrophobic fragmental constants, many biotechnologists have tried since then to correlate effects of organic solvents on biocatalysts they studied with logP approach. Their results have been successful, not completely but only partially. A number of exceptions to the ‘logP rule’ have been in fact reported. 3) Dielectric constant (or dipole moment), ε (or D) Interactions between an enzyme and a solvent in which the enzyme is suspended are mostly non-covalent ones as opposed to interactions in water. These strong non-covalent interactions are essentially of electrostatic origin, and thus according to Coulomb’s law, their strength is imposed dependent on the dielectric constant, ε, (which is higher for water than for almost all organic solvents). It is likely that enzymes are more rigid in anhydrous solvents of low ε than in those of high ε. Thus, ε of a solvent can be used as a criterion of rigidity of the enzyme molecule. For the enzyme to exhibit its activity, it must be dynamically flexible during its whole catalytic action so that its activity in a solvent of lower ε should be less than in a solvent of higher ε. On the other hand, its selectivity or specificity becomes higher when its flexibility decreases so that the selectivity in a solvent of lower ε should be higher than in a solvent of higher ε. 14.4.3.4 Properties of enzymes affected by organic solvents 1) Thermal stability (half-life), t1/2 Stability of an enzyme in an organic solvent is estimated by its half-life, t1/2, when its activity is plotted as a function of incubation time. Although, t1/2 during the enzymatic reaction is more informative for practical purposes, t1/2 under no substrate is often reported because of its easiness of measurement. Inactivation of an enzyme is caused mostly by change in its native conformation, or irreversible unfolding of its native structure. Water, especially enzyme-bound water makes a major contribution to the protein folding through van der Waals interaction, salt-bridges, hydrogen bonds, hydrophobic interaction, etc. When the enzyme molecule is put into organic solvent, water molecules bound to the enzyme molecule are more or less re-equilibrated, depending on the free water content. Therefore, both the nature of organic solvent and the free water content have profound effects on its stability.

876

Tsuneo Yamane

It has been shown that a number of enzymes suspended in anhydrous (dry) organic solvents exhibit thermal stability far superior to that in aqueous solutions (Table 14.4.3.2).11 This is because most of chemical processes that occur in the thermal inactivation involve water, and therefore do not take place in a water-free environment. Furthermore, increased rigidity in dry organic solvents hinders any unfolding process. The increased thermal stability in the dry organic solvent drops down to the stability in aqueous solution by adding small amount of water as demonstrated by Zaks and Klibanov12 and others. Thus, thermal stability of the enzyme in organic solvent strongly depends on its free water content. Table 14.4.3.2. Stability of enzymes in non-aqueous vs. aqueous media. [Adapted, by permission, from ‘Enzymatic Reactions in organic Media’, A. M. P. Koskinen and A. M. Klibanov, Blackie Academic & Professional (An Imprint of Chapman & Hall), Glasgow, 1996, p. 84] Enzyme

Conditions

Thermal property

References

PLL

tributyrin aqueous, pH 7.0

t1/2 < 26 h t1/2 < 2 min

Zaks & Klibanov (1984)

Candida lipase

tributyrin/heptanol aqueous, pH 7.0

t1/2 = 1.5 h t1/2 < 2 min

Zaks & Klibanov (1984)

Chymotrypsin

octane, 100oC aqueous, pH 8.0, 55oC

t1/2 = 80 min t1/2 =15 min

Zaks & Klibanov (1988) Martinek et al. (1977)

Subtilisin

octane, 110oC

t1/2 = 80 min

Russell & Klibanov (1988)

o

Lysozyme

cyclohexane, 110 C aqueous

t1/2 = 140 h t1/2 < 10 min

Ahen & Klibanov (1986)

Ribonuclease

nonane, 110oC, 6 h aqueous, pH 8.0, 90oC

95% activity remains t1/2 < 10 min

Volkin & Klibanov (1990)

F1-ATPase

toluene, 70oC aqueous, 70oC

t1/2 > 24 h t1/2 < 10 min

Garza-Ramos et al. (1989)

Alcohol dehydrogenase

heptane, 55oC

t1/2 > 50 days

Kaul &Mattiasson (1993)

o

HindIII

heptane, 55 C, 30 days

no loss of activity

Kaul & Mattiasson (1993)

Lipoprotein lipase

toluene, 90oC, 400 h

40% activity remains

Ottoline et al. (1992)

β-Glucosidase

2-propanol, 50oC, 30 h

80% activity remains

Tsitsimpikou et al. (1994)

Tyrosinase

chloroform, 50oC aqueous solution, 50oC

t1/2 = 90 min t1/2 = 10 min

Yang & Robb (1993)

Acid phosphatase

hexadecane, 80oC aqueous, 70oC

t1/2 = 8 min t1/2 = 1 min

Toscano et al. (1990)

Cytochrome oxidase

toluene, 0.3% water toluene, 1.3% water

t1/2 = 4.0 h t1/2 = 1.7 min

Ayala et al. (1986)

For References, refer to Ref. 11.


877

2) Specificity and selectivity, kcat/Km It is the most exciting and significant feature that the substrate specificity, enantioselectivity and regioselectivity can be profoundly affected by nature of solvents in which the enzyme molecule exists. This phenomenon has opened an alternative approach for changing specificity and selectivity of an enzyme other than both screening from nature and protein engineering in the field of synthetic organic chemistry. The ability of enzymes to discriminate substrate specificity among different, but structurally similar substrates, enantioselectivity among enantiomers, eantiofaces or identical functional groups linked to a prochiral center, and regioselectivity among identical functional groups on the same molecule, is expressed quantitatively on E value, which is the ratio of the specificity constants, kcat/Km, for the two kinds of substrate (or entiomers), i.e., (kcat/Km)1/(kcat/Km)2. For kinetic resolution of racemic mixture by the enzyme, E is called enantiomeric ratio. The higher E, the higher the enantiomeric excess (i.e., the optical purity), ee, of the product (or remaining substrate). It is said that an E value higher than 100 is preferable for pharmaceutical or biotechnological applications. For overview of this topic, see Refs. 13 and 14. (2a) Substrate specificity Zaks and Klibanov reported that the substrate specificity of α-chymotrypsin, subtilisin, and esterase changed with an organic solvent.15 The substrate specificity in octane was reversed compared to that in Figure 14.4.3.2. The dependence of (A) subtilisin water. A thermodynamical model that preCarlsberg and (B) a-chymotrypsin substrate specificity dicted the substrate specificity of subtilisin for substrates 1 and 2 on the ratio of their Raoult’s law ac- Carlsberg and α-chymotrypsin in organic tivity coefficients. For the structures of the substrates 1 media on the basis of specificity of the enand 2, and the solvents a through m in (A) and a to g in (B), refer to Ref. 16. [Adapted, by permission, from C.R. zyme in water and physicochemical charWescott and A.M. Klibanov, Biotechnol. Bioeng., 56, acteristics of the solvents was developed 343(1997)]. by Wescott and Klibanov.16 They determined kcat/Km for the transesterification of N-acetyl-L-phenylalanine and N-acetyl-L-serine with propanol in 20 anhydrous solvents, and correlated the data of (kcat/Km)Ser/(kcat/Km)Phe, first with the solvent to water partition coefficients for the substrate, PPhe/PSer. Later they examined the selectivity of subtilisin toward two different substrates with the Raoult’s law activity coefficients, γ, by the following equation:15

878

Tsuneo Yamane

log[(k cat / K m )1 / (k cat / K m ) 2 ] = log(γ 1 / γ 2 ) + constant

[14.4.3.3]

The correlation was unexpectedly high as seen in Figure 14.4.3.2,15 implying that the change of substrate specificity of enzyme in organic solvent stems to a large extent from the energy of desolvation of the substrate. (2b) Enantioselectivity Changes in enantioselectivity in various organic solvents was first discovered by Sakurai et al.17 Later Fitzpatrick and Klibanov studied enantioselectivity of subtilisin, Carsberg in the transesterification between the sec-phenethyl alcohol (a chiral alcohol) and vinyl butyrate to find that it was greatly affected by the solvent. Only the correlations with ε or with D gave good agreements.18 The enzyme enantioselectivity was inversed by changing solvents.19 Nakamura et al. studied lipase (Amano AK from Pseudomonas sp.)-catalyzed transesterification of cis- and trans-methylcycolhexanols with vinyl acetate in various organic solvents, and investigated the effect of solvent on activity and stereoselectivity of the lipase.20 They correlated their stereoselectivity with good linearity (except for dioxane and dibutyl ether) by the following two-parameter equation (Figure 14.4.3.3): E = a(ε − 1) / (2ε + 1) + bVm + c

[14.4.3.4]

where ε and Vm are dielectric constant and molar volume of the solvent, respectively, and a, b, and c are constants which should be experimentally determined. Bianchi et al. also reported that for the resolution of antitussive agent, Dropropizine, using both hydrolysis in aqueous buffer and transesterification techniques in various organic solvents, by a lipase (Amano PS from Pseudomonas cepacia), E depended very much on organic solvents (Table 14.4.3.3), with the highest E value (589) in n-amyl alcohol and the lowest one (17) in water.21 In this case, however, there was no correlation between the enantioselectivity and the Figure 14.4.3.3. Linear relationship between f(ε, Vm) and E for physico-chemical properties of lipase-catalyzed transesterification of cis- and trans-4-methylthe solvents such as logP or ε. cyclohexanols with vinyl acetate in various organic solvents. For the organic solvents 1 through 18, refer to Ref. 20. [Adapted, by permission, from K. Nakamura, M. Kinoshita and A. Ohno, Tetrahedron, 50, 4686(1994)].


879

Table 14.4.3.3. Effect of the solvent on enantioselectivity of lipase PS. [Adapted, by permission, from D. Bianchi, A. Bosetti, P. Cesti and P. Golini, Tetrahedron Lett., 33, 3233(1992)]. Solvent

logP

H 2O

ε

Nucleophile

E

78.54

H2O

17

hexane

3.5

1.89

n-propanol

146

CCl4

3.0

2.24

n-propanol

502

toluene

2.5

2.37

n-propanol

120

iso-propyl ether

1.9

3.88

n-propanol

152

2-methyl-2-butanol

1.45

5.82

n-propanol

589

2-methyl-2-butanol

1.45

5.82

H2O

63

n-propanol

0.28

20.1

n-propanol

181

acetonitrile

-0.33

36.2

n-propanol

82

1,4-dioxane

-1.1

2.2

n-propanol

164

As mentioned above, no correlation was reported in a large number of articles on the effects of solvents on the enzyme specificity/selectivity, although correlations between the specificity/selectivity and physico-chemical properties of the solvents were successful in some combinations of an enzymatic reaction for a set of solvents. Therefore, attempts to rationalize the phenomena based on either physico-chemical properties of the solvents or on their structure, are at present clearly unsatisfactory from the point of view of predictable value. Further experiments carried out under more strictly defined condition are necessary to reach the quantitative explanation of the whole phenomena. 14.4.3.5 Concluding remarks Activity, stability, and selectivity of an enzyme are affected considerably by nature of organic solvents as well as free water content in the enzyme-catalyzed reaction in the organic solvents. However, rational criteria for the selection of a proper solvent among vast variety of the organic solvents are very limited so far. Researchers are obliged to resort to empirical approach by examining some kinds of solvent for the enzyme-catalyzed reaction concerned at the present state of art. References 1 2 3 4 5 6 7 8 9 10

A.M.P. Koskinen and A.M. Klibasnov, Ed., Enzymatic Reactions in Organic Media, Blackie Academic & Professional (An Imprint of Chapman & Hall), London, 1996. T. Yamane, Nippon Nogeikagaku Kaishi, 65, 1103(1991). T. Yamane, J. Am. Oil Chem. Soc., 64, 1657(1987). T. Yamane, Y. Kozima, T. Ichiryu and S. Shimizu, Ann. New York Acad. Sci., 542, 282(1988). T. Yamane, Biocatalysis, 2, 1(1988). J.A. Rapley, E. Gratton and G. Careri, Trends Biol. Sci., Jan., 18(1983). P.J. Halling, Biochim. Biophys. Acta, 1040, 225(1990). R.H. Valivety, P.J. Halling, A.R. Macrae, Biochim. Biophys. Acta, 1118, 218(1992). C. Laane, S. Boeren and K. Vos, Trends Biotechnol., 3, 251(1985). C. Laane, S. Boeren, K. Vos and C. Veeger, Biotechnol. Bioeng., 30, 81(1987).

880

11 12 13 14 15 16 17 18 19 20 21

George Wypych A. Zaks, in Enzymatic Reactions in Organic Media ed. by A.M.P. Koskinen and A.M. Klibanov, Blackie Academic & Professional (An Imprint of Chapman & Hall), London, 1996, p. 84. A. Zaks and A.M. Klibanov, Science, 224, 1249(1984). C.R. Wescott and A.M. Klibanov, Biocheim. Biophys. Acta, 1206, 1(1994). G. Carrea, G. Ottolina and S. Riva, Trends Biotechnol., 13, 63(1995). A. Zaks and A.M. Klibanov, J. Am. Chem. Soc., 108, 2767(1986). C.R. Wescott and A.M. Klibanov, J. Am. Chem. Soc., 108, 2767(1986). T. Sakurai, A.L. Margolin, A.J. Russell and A.M. Klibanov, J. Am. Chem. Soc., 110, 7236(1988). P.A. Fitzpatrick and A.M. Klibanov, J. Am. Chem. Soc., 113, 3166(1991). S. Tawaki and A.M. Klibanov, J. Am. Chem. Soc., 114, 1882(1992). K. Nakamura, M. Kinoshita and A. Ohno, Terehedron, 50, 4681(1994). D. Bianchi, A. Bosetti, P. Cesti and P. Golini, Tetrahedron Lett., 33, 3231(1992).

14.5 COIL COATING George Wypych ChemTec Laboratories, Inc., Toronto, Canada

The coil coating industry is under pressure to eliminate the use of solvents. Polyester coil coatings contain up to 40% of solvents such as glycol esters, aromatic hydrocarbons (e.g., Solvesso 150), alcohols, ketones, and butyl glycol.1,2 A recent book1 predicted that the solvent-based technology will not change during the next decade because the industry heavily invested in equipment to deal with solvents. Such changes in technology require long testing before they can be implemented. The coil coating industry normally recovers energy from evaporated solvents either by at-source incineration or by a recycling process which lowers emissions. Because of the large amount of solvents used, the use of PVC and fluorocarbon resins in some formulations, and the use of chromates in pretreatments the pressure remains on the industry to make improvements.3 The coil coating industry is estimated to be consuming about 50,000 tons of solvents both in Europe and in the USA.1 About half of these solvents are hydrocarbons. According to the published studies,3,4 efforts to change this situation did start in the early 1990s and by mid nineties research data were available to show that the technology can be changed. Two directions will most likely challenge the current technology: radiation curing and powder coating. Coil coats are thin (about 30 µm wet thickness) but contain a high pigment loading. Consequently, UV curing is less suitable than electron beam curing. The application of this technology requires a change to the polymer system and acrylic oligomers are the most suitable for this application. This system can be processed without solvents. If a reduction of viscosity is required, it can be accomplished by the use of plasticizers (the best candidates are branched phthalates and linear adipates) and/or reactive diluents such as multifunctional monomers. Results3 show that the UV stability of the system needs to be improved by using a polyester top coat or fluoropolymer. With top coat, material performs very well as learned from laboratory exposures and exposures in industrial environment.3 At the time of the study (about 6 years ago), process of coating was less efficient than solvent-based system because production speed was about 6 times slower than the highest production rates in the industry (120 m/min). At the same time, it is known2 that the quality of solvent-based coatings suffers from excessive production rates as well. Radiation curing has a disadvantage because of its high capital investment but it does have an economical advantage because the

14.6 Cosmetics and personal care products

881

process is very energy efficient. Previous experiences with radiation curing technology show that the process has been successfully implemented in several industries such as paper, plastic processing, and wood coating where long term economic gains made the changes viable. Comparison of solvent-based fluoropolymer and fluorocarbon powder coating developed in Japan4 shows that elimination of solvent is not only good for environment but also improves performance (UV stability especially is improved). The study was carried out with a very well designed testing program to evaluate the weathering performance of the material. These two technologies show that there is extensive activity to improve coil coatings with simultaneous elimination of solvents. Two recent patents contribute more information on the developments in the coil coating industry.5,6 One problem in the industry is with the poor adhesion of the coating to steel.2,3 A primer developed contains dipropylene glycol methyl ether and PM acetate which allows the deposition of relatively thick layers (20-40 µm) without blistering and at suitable rate of processing. However, the primer has a low solids content (30-45%).5 A new retroreflective coating was also developed6 which is based on ethyl acrylate-styrene copolymer and contains a mixture of xylene with another aromatic hydrocarbon (Solvesso 150) at relatively low concentration (11-12%).

REFERENCES 1 2 3 4 5 6

B P Whim, P G Johnson, Directory of Solvents, Blackie Academic & Professional, London, 1996. A L Perou, J M Vergnaud, Polym. Testing, 16, No.1, 19-31 (1997). G M Miasik, Surface Coatings International, 79, No.6, 258-67 (1996). C Sagawa, T Suzuki, T Tsujita, K Maeda, S Okamoto, Surface Coatings International, 78, No.3, 94-8 (1995). M T Keck, R J Lewarchik, J C Allman, US Patent 5,688,598, Morton International, Inc., 1997. G L Crocker, R L Beam, US Patent 5,736,602, 1998.

14.6 COSMETICS AND PERSONAL CARE PRODUCTS George Wypych ChemTec Laboratories, Inc., Toronto, Canada Several cosmetic products contain solvents. These include nail polish, nail polish remover, fragrances, hair dyes, general cleaners, hair sprays and setting lotions. In most cases, ethanol is the only solvent. Nail polish and nail polish remover contain a large variety of solvents. Several recent patents3-6 give information on current developmental work in this area. Nitrocellulose, polyester, acrylic and methacrylic ester copolymer, formaldehyde resin, rosin, cellulose acetate butyrate are the most frequently used polymers in nail polish formulations. Solvents were selected to suit the polymer used. These include acetone, methyl acetate, ethyl acetate, butyl acetate, methyl glycol acetate, methyl ethyl ketone, methyl isobutyl ketone, toluene, xylene, isopropyl alcohol, methyl chloroform, and naphtha. Solvents constitute a substantial fraction of the composition usually around 70%. Reformulation is ongoing to improve the flexibility and durability of the nail polish.3 Other efforts are directed to improve antifungal properties,4 to eliminate ketones and formaldehyde resin (ketones because of their toxicity and irritating smell and formaldehyde resins because they contribute to dermatitis),5 and elimination of yellowing.6 All efforts are di-

882

George Wypych

rected towards improvements in drying properties and adhesion to the nails. These properties are partially influenced by solvent selection. The current trend is to greater use of ethyl acetate, butyl acetate, naphtha, and isopropanol which are preferable combinations to the solvents listed above. Acetone used to be the sole component of many nail polish removers. It is still in use but there is a current effort to eliminate the use of ketones in nail polish removers. The combinations used most frequently are isopropanol/ethyl acetate and ethyl acetate/isopropanol/ 1,3-butanediol. General cleaners used in hairdressing salons contain isopropanol and ethanol. Hair spray contains ethanol and propellants which are mixtures of ethane, propane, isobutane, and butane. The reported study2 of chemical exposure in hairdresser salons found that although there were high concentrations of ethanol the detected levels were still below the NIOSH limit. The concentrations were substantially higher in non-ventilated salons (about 3 times higher) than those measured in well ventilated salons. Small concentrations of toluene were found as well, probably coming from dye components. Recent patents7,8 show that solvents may enter cosmetic products from other ingredients, such as components of powders and thickening agents. Some solvents such as dichloromethane and benzene, even though they are present is smaller quantities are reason for concern.

REFERENCES 1 2 3 4 5 6 7 8

B P Whim, P G Johnson, Directory of Solvents, Blackie Academic & Professional, London, 1996. B E Hollund, B E Moen, Ann. Occup. Hyg., 42 (4), 277-281 (1998). M F Sojka, US Patent 5,374,674, Dow Corning Corporation. 1994. M Nimni, US Patent 5,487,776, 1996. F L Martin, US Patent 5,662,891, Almell Ltd., 1997. S J Sirdesai, G Schaeffer, US Patent 5,785,958, OPI Products, Inc., 1998. R S Rebre, C Collete, T. Guertin, US Patent 5,563,218, Elf Atochem S. A., 1996. A Bresciani, US Patent 5,342,911, 3V Inc., 1994.

14.7 Dry cleaning

883

14.7 DRY CLEANING - TREATMENT OF TEXTILES IN SOLVENTS Kaspar D. Hasenclever Kreussler & Co.GmbH, Wiesbaden, Germany

Most processes in manufacturing and finishing of textiles are aqueous. In order to prevent water pollution, some years ago developments were made to transform dyeing-, cleaningand finishing-processes from water to solvents. During 1970-1980 solvent processes for degreasing, milling, dyeing and waterproofing of textiles could get limited economic importance. All these processes were done with tetrachloroethylene (TCE). After getting knowledge about the quality of TCE penetrating solid floors, stone and ground, contaminating groundwater, this technology was stopped. Today the main importance of solvents in connection with textiles is given to dry cleaning, spotting and some special textile finishing processes. 14.7.1 DRY CLEANING 14.7.1.1 History of dry cleaning Development of dry cleaning solvents The exact date of discovery of dry cleaning is not known. An anecdote tells us that in about 1820 in Paris, a lamp filled with turpentine fell down by accident and wetted a textile. After the turpentine was vaporized, the wetted areas of the textile were clean, because the turpentine dissolved oily and greasy stains from it. In 1825, Jolly Belin founded the first commercial “dry laundry” in Paris. He soaked textile apparel in a wooden tub filled with turpentine, cleaned them by manual mechanical action, and dried them by evaporating the turpentine in the air. After getting the know-how to distill benzene from tar of hard coal in 1849, this was used as a solvent for dry cleaning because of its far better cleaning power. But benzene is a strong poison, so it was changed some decades later to petrol, which is explosive. In order to reduce this risk, petrol as dry cleaning solvent was changed to white spirit (USA: Stoddard solvent) with a flash point of 40 - 60°C (100 - 140°F) in 1925. The flammability of the hydrocarbon solvents in dry cleaning plants was judged to be risky because of fire accidents. After finding the technology for producing inflammable chlorinated hydrocarbons, trichloroethane and tetrachloroethylene (TCE) were introduced in dry cleaning since about 1925. These solvents gave the opportunity for good cleaning results and economic handling. Up to 1980, TCE was the most important solvent for dry cleaning worldwide. Compared to TCE, fluorinated chlorinated hydrocarbons (CFC) offer benefits to dry cleaning because of their lower boiling points and their more gentle action to dyestuffs and fabrics. So since 1960 these solvents have had some importance in North America, Western Europe, and the Far East. They were banned because of their influence on the ozone layer in the stratosphere by the UNESCO’s Montreal Protocol in 1985. At the same time, TCE was classified as a contaminant to groundwater and as a dangerous chemical to human health with the possible potential of cancerogenic properties. As a result of this, hydrocarbon solvents on the basis of isoparaffins with a flash point higher

884

Kaspar D. Hasenclever

than 55°C (130°F) were at first used in Japanese and German dry cleaning operations since 1990. Also, alternatives to conventional dry cleaning were developed. Wet cleaning was introduced by Kreussler in 1991 (Miele System Kreussler) and textile cleaning in liquid carbon dioxide was exhibited by Global Technologies at the Clean Show Las Vegas in 1997. Development of dry cleaning machines In the beginning, dry cleaning was done manually in wooden tubs filled with turpentine or benzene. About 1860, the Frenchman Petitdidier developed a wooden cylindrical cage, which was rotated in a tub filled with solvent. The apparel to be dry cleaned was brought into the cage and moved through the solvent by the rotation of the cylinder. This machine got the name “La Turbulente”. The next step was the addition of a centrifuge to the wooden machine. The dry cleaned apparel was transferred from the machine into the centrifuge and then dried by vaporizing the solvent in the open air. About 1920, a tumble dryer was used for drying the dry cleaned textiles. Fresh air was heated up, blown through the dryer, where the air was saturated with solvent vapor and then blown out to be exhausted into the environment. The used solvent was cleaned to be recycled by separation of solid matter by centrifugal power with a separator and to be cleaned from dissolved contamination by distilling. In 1950, dry to dry machines were developed by Wacker in Germany for use with TCE. The principle of working is as follows: The cage, filled with textiles, is rotating in a closed steel cylinder. The dry cleaning solvent is pumped from the storage tank into the cylinder so that the textiles in the cage are swimming in the solvent. After ending the cleaning process, the solvent is pumped back into the storage tank and the cage rotates with high speed (spinning) in order to separate the rest of the solvent from the textile. Then air is circulated through a heat exchanger to be heated up. This hot air is blown into the rotating cage in order to vaporize the remaining solvent from the textiles. The air saturated with solvent vapor is cleaned in a condenser, where the solvent is condensed and separated from the air. The air now goes back into the heat exchanger to be heated up again. This circulation continues until textiles are dry. The separated solvent is collected in a tank to be reused. In order to reduce solvent losses and solvent emissions, since 1970 charcoal filters have been used in the drying cycle of dry cleaning machines, so that modern dry cleaning systems are separated from the surrounding air. 14.7.1.2 Basis of dry cleaning Dry cleaning means a cleaning process for textiles, which is done in apolar solvents instead of water. If water is used, such cleaning process is called “washing” or “laundering”. Natural textile fibers, such as wool, cotton, silk and linen, swell in water because of their tendency to absorb water molecules in themselves. This causes an increase of their diameter and a change of the surface of yarns and fabrics. The result is shrinkage, felting and creasing. Apolar solvents, such as hydrocarbons, are not absorbed by natural textile fibers because of the high polarity of the fibers. So there is no swelling, no shrinkage, no felting and no creasing. From the solvent activity, dry cleaning is very gentle to textiles, with the result that the risk of damaging garments is very low. Because of the apolar character of dry cleaning solvents cleaning activity also deals with apolar contamination. Oils, fats, grease and other similar substances are dissolved in

14.7 Dry cleaning

885

the dry cleaning operation. Polar contamination, such as, salt, sugar, and most nutrition and body excrements are not dissolved. In the same way that water has no cleaning activity with regard to oils, fat, or grease, in dry cleaning there is no activity with regard to salts, sugar, nutrition and body excrements. And in the same way that water dissolves these polar substances, cleaning of textiles from these contaminants does not pose any problems in a washing process. To make washing active to clean oils, fat and grease from textiles, soap (detergent) has to be added to water. To make dry cleaning active to clean salts, sugar and the like from textiles, dry cleaning detergent has to be added to the solvent. In washing, fresh water is used for the process. After being used, the dirty washing liquid is drained off. In dry cleaning the solvent is stored in a tank. To be used for cleaning it is pumped into the dry cleaning machine. After being used, the solvent is pumped back into the storage tank. To keep the solvent clean it is constantly filtered during the cleaning time. In addition to this, a part of the solvent is pumped into a distilling vessel after each batch to be cleaned by distilling. Dry cleaning solvents are recycled. The solvent consumption in modern machines is in the range of about 1 - 2 % per weight of the dry cleaned textiles. 14.7.1.3 Behavior of textiles in solvents and water Fibers used for manufacturing textiles can be classified into three main groups: • Cellulosic fibers: cotton, linen, rayon, acetate. • Albumin fibers: wool, silk, mohair, camelhair, cashmere. • Synthetic fibers: polyamide, polyester, acrylic. Textiles made from cellulosic fibers and synthetics can be washed without problems. Apparel and higher class garments are made from wool and silk. Washing very often bears a high risk. So these kinds of textiles are typically dry cleaned. Dependent to the relative moisture of the surrounding air fabrics absorb different quantities of water. The higher the polarity of the the fiber, the higher is their moisture content. The higher the swelling (% increase of fiber diameter) under moisture influence, the higher is the tendency of shrinkage in a washing or dry cleaning process. Table 14.7.1. Water content (%) in textile fibers dependent on relative humidity

Fiber

Relative humidity, %

Swelling, %

70

90

max.

Viscose

14.1

23.5

24.8

115

Wool

15.6

22.2

28.7

39

Silk

11.2

16.2

17.7

31

Cotton

8.1

11.8

12.9

43

Acetate

5.4

8.5

9.3

62

Polyamide

5.1

7.5

8.5

11

Acrylic

2.1

4.0

4.8

9

Polyester

0.5

0.6

0.7

0

886


The same absorption of water occurs if textiles are immersed in solvent in a dry cleaning machine. If the relative humidity in the air space of the cylinder of a dry cleaning machine is higher, textiles absorb more water. The water content of textiles in solvent is equal to the water content of textiles in the open air if the relative humidity is the same. If water additions are given to the dry cleaning solvent in order to intensify the cleaning effect with regard to polar contamination, then the relative humidity in the air space of the cylinder of the dry cleaning machine increases. As a result of this, the water content in the dry cleaned textiles increases, too, so that swelling begins and shrinkage may occur. Woolen fabrics are particularly sensitive to shrinkage and felting because of the scales on the surface of the wool fiber. Not only in washing, but also in dry cleaning there is a risk of shrinkage on woolen garments. This risk is higher when the dry cleaning process is influenced by water addition. If the relative humidity in the air space of a dry cleaning machine is more than 70%, shrinkage and felting may occur, if the dry cleaning solvent is tetrachloroethylene (TCE). In hydrocarbon solvents wool is safe up to 80% relative humidity, because of the lower density of hydrocarbons compared to TCE, which reduces mechanical action. 14.7.1.4 Removal of soiling in dry cleaning “Soiling” means all the contamination on textiles during their use. This contamination is of very different sources. For cleaning purposes, the easiest way of classification of “soiling” is by solubility of soiling components. The classification can be done by definition of four groups: • Pigments: Substances are insoluble in water and in solvents. Examples are: dust, particles of stone, metal, rubber; soot, scale of skin. • Water soluble material: Examples are: salts, sugar, body excrements, sap and juice. • Polymers: Substances are insoluble in solvents but can be soaked and swell in water. Examples are: starch, albumin and those containing material such as blood, milk, eggs, sauce. • Solvent soluble material: Examples are: oil, fat, grease, wax, resins. Table 14.7.2 Average soiling of garments (apparel) in Europe Soil type

Proportion, %

Solubility

Components

Pigments

50

not

Water soluble

30

water

sugar, salt, drinks, body excretions

Polymers

10

water

starch, albumin, milk, food

Solvent soluble

10

solvents

dust, soot, metal oxides, rub-off, pollen, aerosols

skin grease, resin, wax, oils, fats

In practice, the situation is not so simple as it seems to appear after this classification. That is because soiling on garments almost always contains a mixture of different substances. For example, a spot of motor oil on a pair of trousers consists of solvent soluble oil, but also pigments of soot, metal oxides and other particles. The oil works as an adhesive for

14.7 Dry cleaning

887

the pigments, binding them to fabric. In order to remove this spot, first the oil has to be dissolved then pigments can be removed. The removal of solvent soluble “soiling” in dry cleaning is very simple. It will be just dissolved by physical action. Polymers can be removed by the combined activity of detergents, water and mechanical action. The efficiency of this process depends on the quality and concentration of detergent, the amount of water added into the system, and the operating time. Higher water additions and longer operating time increase the risk of shrinkage and felting of the textiles. Water soluble “soiling” can be removed by water, emulsified in the solvent. The efficiency here depends on the emulsifying character of the detergent and the amount of water addition. The more water is emulsified in the solvent, the higher is the efficiency of the process and the higher is the risk of shrinkage and felting of the textiles. Pigments can be removed by mechanical action and by dispersing activity of detergents. The higher the intensity of mechanical action, influenced by cage diameter, rotation, gravity of the solvent and operating time and the better the dispersing activity of the detergent, the better is the removal of pigments. The same parameters influence the care of the textiles. The better the cleaning efficiency, the higher the risk for textile damage. 14.7.1.5 Activity of detergents in dry cleaning The main component of detergents is surfactant. The eldest known surfactant is soap. Chemically soap is an alkaline salt of fatty acid. Characteristic of soap (and surfactant) is the molecular structure consisting of apolar - hydrophobic - part (fatty acid) and a polar - hydrophilic - part (-COONa), causing surface activity in aqueous solution. Surface activity has its function in the insolubility of the hydrophobic part of molecule in water and the hydrophilic part of molecule influences water solubility. This gives a tension within soap molecules in water forming layers on every available surface and forming micelles if there is a surplus of soap molecules compared to the available surface. This soap behavior stands as an example of mechanism of action of surfactants in general. Micelles of surfactants in water are formed by molecular aggregates of surfactants oriented in such a way that the hydrophobic parts are directed internally, so that the hydrophilic parts are directed outwards. In this way the aggregates form spheres, cylinders, or laminar layers, dependent on its concentration. Because of this behavior, it is possible to remove oil, fat or grease from substrates in aqueous solutions, if surfactants are present. The surfactants act to disperse the oil into small particles and build up micelles around these particles, so that oil, fat or grease incorporated inside the micelle (Figure 14.7.1). If aggregates are small, the solution is clear. If aggregates are larger, the solution (emulsion - type oil in water) becomes milky. In the same way, but in the opposite direction, surfactants form micelles in solvents (Figure 14.7.2). In this case, not the hydrophobic, but the hydrophilic part is directed internally and the hydrophobic outwards. Figure 14.7.2. Schematic Figure 14.7.1. Schematic Emulsions in this case are not formed by oil diagram of surfactants agdiagram of surfactants aggregated in micelle in solgregated in micelle in in water, but by water in oil (solvent). aqueous solution.

vent solution.

888


Because of this behavior, it is possible to remove water soluble (polar) material from substrates in solvents, if surfactants are at present. The surfactants disperse the polar material into very small particles and build up micelles around the particles, so that the polar material is totally incorporated inside micelle. If the aggregates are small, the solution is clear. If they are large, the solution (emulsion type water in oil) becomes milky. Drycleaning detergents consist of surfactants, cosolvents, lubricants, antistatic compounds and Figure 14.7.3. Aggregation of surfactants in emulsions of type oil in water (O/W) and water water. They are used in order to increase the cleaning efficiency, to improve the handle of dry cleaned in oil (W/O). textiles, and to prevent electrostatic charge on textiles. They are formulated as liquids in order to be easily added to the solvent by automatic dosing equipment. In dry cleaning the drycleaning detergents play the same role as soap or laundry detergents in textile washing. 14.7.1.6 Dry cleaning processes Process technology in dry cleaning has the target to clean garments as good as possible without damaging them, at lowest possible costs and with highest possible safety. Because of the environmental risks dependent on the use of TCE or hydrocarbon solvents, safe operation is the most important. Modern dry cleaning machines are hermetically enclosed, preventing infiltration of surrounding air, operating with electronic sensor systems, and they are computer controlled. The main part is the cylinder with the cleaning cage, the solvent storage tanks, the solvent recovering system, the solvent filter system and the distillation equipment (Figure 4.9.4). Cylinder/cage: A cage capacity is 20 liters per kg load capacity of the machine (about 3 gal per lb). That means, an average sized dry cleaning machine with load capacity of 25 kg (50 lbs) has a cage capacity of 500 liters (130 gal). This size is necessary because the same cage is used for cleaning and drying. During the cleaning cycle, 3 - 5 liters of solvent per kg load (0.5 - 0.8 gal/lb) are pumped into the cylinder. In a 25 kg machine, it is 80 - 125 liters (25-40 gal). This solvent is filtered during the cleaning time. Solvent storage tanks: A dry cleaning machine has 2 - 3 solvent storage tanks. The biggest - the working tank - has a capacity in liters ten times the load capacity in kg. The clean solvent tank and optional retex tank have half the capacity. All the tanks are connected to each other with an overflow pipe. The working tank with its inlet and outlet is connected to the cylinder of the machine. The inlet of the clean tank is connected to the distilling equipment, the outlet to the cylinder. Filter system: The filter is fed with solvent from the cylinder by pump pressure. The filtered solvent goes back into the cylinder. The filter has a further connection to the distilling equipment used as drain for the residue.

14.7 Dry cleaning

889

Figure 14.7.4. Scheme of a dry cleaning machine. 1 textiles to be cleaned, 2 cylinder, 3 cage, 4 fan, 5 heater, 6 drive for cage, 7 air canal, 8 cooler, 9 lint filter, 10 button trap, 11 filter, 12 distilling vessel, 13 pump, 14 solvent storage tanks, 15 water separator, 16 condenser, 17 steam pipe, 18 outlet to charcoal filter, 19 dosing unit.

Distilling equipment: The distilling vessel has the same capacity as the biggest storage tank. Solvent to be distilled is pumped into the vessel from the cylinder or from the filter. The vessel is steam or electrical heated. The solvent vapor is directed into a condenser (water cooling) and then to a water separator, from where it flows into the clean solvent tank. The residue after distillation remains in the vessel and is pumped into special residue drums. Recovering system: In order to dry the dry cleaned garments from the solvent residue which remains after spinning, a fan extracts air out of the cylinder into a condenser, where solvent vapor is condensed out of the air. From there the air is blown into a heater and directed by fan back into the cylinder. In this way the solvent is vaporized away from the garment and after condensing, reused for cleaning. Dosing equipment: In order to get the right additions of dry cleaning detergent into the solvent, a dry cleaning machine has dosing equipment working on the basis of a piston pump, which doses the right amount of detergent at the right time into the system.

890


Computer control: All the processes are computer controlled. Drying/Recovering, Distilling/Filtration operate according to fixed programs; the cleaning cycle can be varied with regard to the requirements of the garment. The most common program works as follows: Loading the machine, closing the door, starting computer control: 1 min pumping solvent from working tank into cylinder up to low dip level (3 l/kg); 3 min pre-cleaning by rotating the cage in the cylinder; 1 min pumping the solvent from the cylinder into the distillation vessel; 1 min spinning; 2 min pumping solvent from working tank into cylinder up to high dip level (5 l/kg); addition of detergent (2-5 ml/l of solvent); 8 min cleaning by rotating the cage in the cylinder together with filter action; 1 min pumping the solvent from the cylinder back to the working tank (the solvent in working tank is filled by overflow from clean tank); 3 min spinning 8 min drying /recovering solvent with recovery system 5 min drying/recovering solvent with charcoal filter. After the sensor has indicated that the load is free of solvent residue, opening the door and unloading the machine. After the cleaning process, the garment is controlled for cleaning quality and can be finished or if necessary, it undergoes spotting/recleaning before finishing. 14.7.1.7 Recycling of solvents in dry cleaning The recycling of solvents in dry cleaning is very important, because solvents are too expensive for single use. Three different systems are used for the cleaning of the solvent in order for it to be recycled: • Filtration • Adsorption • Distillation Filtration is a simple physical process, separating insoluble parts from the solvent. It is done during the cleaning cycle. Adsorption is mainly used together with hydrocarbon solvents, because their high boiling temperature is insufficient to separate lower boiling contaminants from the solvent. Adsorption systems use charcoal or bentonites. The solvent is pumped to filters where contaminants with higher polarity than solvent are adsorbed by the adsorbing material. The adsorbing material can adsorb contaminants in a quantity of about 20% of its own weight. After being saturated, the adsorbing material must be replaced and changed to fresh material. The charged adsorbing material is disposed according to regulations, which is cost intensive. Distillation is the best cleaning method if the boiling point of solvent is significantly lower than the boiling point of possible contaminants. With dry cleaning machines using TCE, distilling is the normal recycling method. The boiling point of TCE is 122°C, which makes steam heating possible, so that the process can be done safely and cost effectively. The distillation residue consists of removed soil and detergents. Its quantity is much lower

14.7 Dry cleaning

891

than in the adsorption systems, because there is no waste from adsorbing material. Disposal costs are much lower than with adsorption. 14.7.2 SPOTTING 14.7.2.1 Spotting in dry cleaning Spotting is the removal of stains from textiles during professional textile cleaning. Correct cleaning, good lighting conditions, appropriate equipment, effective spotting agents, and expert knowledge are indispensable. Lighting conditions Lighting of the spotting table with composite artificial light consisting of a bluish and a yellowish fluorescent tube attached above the table approx. 80 cm from the standing point is better suited than daylight. This ensures high-contrast, shadow-free lighting of the work surface and allows for working without fatigue. Equipment Basic equipment should comprise a spotting table with vacuum facility, sleeve board, steam and compressed air guns, and a spray gun for water. Spotting brushes should have soft bristles for gentle treatment of textiles. Use brushes with bright bristles for bright textiles and brushes with dark bristles for dark textiles. Use spatulas with rounded edges for removing substantial staining. 14.7.2.2 Spotting agents In dry cleaning, normally three groups of spotting agents are used: Brushing agents Brushing agents containing surfactants and glycol ethers dissolved in low viscosity mineral oil and water. Brushing agents are used for pre-spotting to remove large stains from textiles. They are applied undiluted with a soft brush or sprayed onto the heavily stained areas before dry cleaning. Special spotting agents Special spotting agents are used for removing particular stains from textiles. The range consists of three different products in order to cover a wide range of different stains. The products are applied as drops directly from special spotting bottles onto the stain and are allowed to react. The three products are: • Acidic solution of citric acid, glycerol surfactants, alcohol and water for removing stains originating from tannin, tanning agent, and fruit dye. • Basic solution of ammonia, enzymes, surfactants, glycol ethers and water for removing stains originating from blood, albumin, starch, and pigments. • Neutral solution of esters, glycol ethers, hydrocarbon solvents and surfactants for removing stains originating from paint, lacquer, resin, and adhesives. Post-spotting agents Stains that could not be removed during basic cleaning must be treated with post-spotting agents. Most common is a range of six products, which are used in the same way, as the special spotting agents: • Alkaline spotting agent for stains originating from starch, albumin, blood, pigments • Neutral spotting agent for stains originating from paint, lacquer, grease, and make-up. • Acidic spotting agent for stains originating from tannin, fruits, beverages, and rust. • Acidic rust remover without hydrofluoric acid.

892


• Solvent combination for removing oily and greasy stains from textiles. • Bleaching percarbonate as color and ink remover. 14.7.2.3 Spotting procedure Correct procedure for spotting It is recommended to integrate the three stages of spotting • Brushing • Special spotting • Post-spotting into the process of textile cleaning. Perform brushing and special spotting when examining and sorting the textiles to be cleaned. Brushing Examine textiles for excessive dirt, particularly at collars, pockets, sleeves, and trouser legs. If textiles are to be dry cleaned check them particularly for stains originating from food or body secretions. If the textiles are to be wet-cleaned check them particularly for greasy stains. Apply a small quantity of brushing agent onto the stained areas and allow to react for 10-20 minutes before loading the cleaning equipment. Special spotting Intensive staining found during the examination of the textiles can be treated with special spotting agents. The stain substance must be identified and related to one of the following categories: • coffee, tea, fruit, red wine, grass, urine • blood, food, pigments, sweat • wax, paint, lacquer, make-up, pen ink, adhesives Depending on the category of the stain substance, apply the special product with the dripping spouts in the work bottle onto the stain and tamp it gently with a soft spotting brush. Allow to react for 10-20 minutes before loading the cleaning equipment. Post-spotting Stains that could not be removed during dry-cleaning or wet cleaning with machines are subject to post-spotting. Proceed as follows: • Place the garment with the stain area onto the perforated vacuum surface of the spotting table. • Identify stain, drip appropriate product undiluted onto the stain and tamp it gently with the brush. • For stubborn stains, allow product to react for up to 3 minutes. • Use vacuum to remove product and rinse spotted area with steam gun. • Dry with compressed air, moving the air gun from the edges to the center of the spotted area. Hidden spot test If it is suspected that a stain cannot be removed safely due to the textile material, the compatibility of the spotting agent should be checked by applying a small quantity of the agent at a hidden part of the garment. If the garment passes the test, the spotting agent is expected to be successful without damaging the garment.

14.7 Dry cleaning

893

14.7.3 TEXTILE FINISHING The use of solvents for textile finishing has only some importance for the treatment of fully fashioned articles. The processes are done in industrial dry cleaning machines. The advantages are the same as in dry cleaning compared to washing: lower risk against shrinkage and damage of sensitive garments. 14.7.3.1 Waterproofing In Northern Europe, North America and Japan, some kinds of sportsware need to be waterproofed. The treatment is done in dry cleaning machines with a load capacity larger than 30 kg ( > 60 lbs). The machines need to be equipped with a special spraying unit, which allows one to spray a solution of waterproofing agent into the cage of the machine. The waterproofing agents consists of fluorocarbon resins dissolved in a mixture of glycol ethers, hydrocarbon solvents and TCE. This solution is sprayed onto the garments, which are brought into the cage of the machine. The spraying process needs about 5 - 10 min. After spraying, the solvent is vaporized in the same way as drying in dry cleaning, so that the fluorocarbon resin will stay on the fibers of the garments. In order to get good results and the highest possible permanence of water resistance, the resin needs to be thermally fixed. In order to meet these requirements, a drying temperature of > 80°C (= 175°F) is necessary. 14.7.3.2 Milling Solvent milling has some importance for the treatment of fully fashioned woolen knitwear. The process is very similar to normal dry cleaning. The specific difference is the addition of water together with the detergent, in order to force an exact degree of shrinkage and/or felting. Milling agents are similar to dry cleaning detergent. They have specific emulsifying behavior, but no cleaning efficiency. The process runs like this: textiles are loaded into the machine, then solvent (TCE) is filled in before the milling agent diluted with water is added. After this addition, the cage rotates for 10 - 20 min. The higher the water addition, the higher the shrinkage; the longer the process time, the higher the felting. After this treatment, the solvent is distilled and textile load is dried. High drying temperature causes a rather stiff handle, low drying temperatures give more elastic handle. 14.7.3.3 Antistatic finishing Antistatic finishing is used for fully fashioned knitwear - pullovers made from wool or mixtures of wool and acrylic. The process is equal to dry cleaning. Instead of dry cleaning detergent, the antistatic agent is added. Antistatic agents for the treatment of wool consist of cationic surfactants such as dialkyl-dimethylammonium chloride, imidazolidione or etherquats. Antistatic agents for the treatment of acrylic fibers are based on phosphoric acid esters.

894

Martin Hanek, Norbert Löw, Andreas Mühlbauer

14.8 ELECTRONIC INDUSTRY - CFC-FREE ALTERNATIVES FOR CLEANING IN ELECTRONIC INDUSTRY Martin Hanek, Norbert Löw Dr. O. K. Wack Chemie, Ingolstadt, Germany

Andreas Mühlbauer Zestron Corporation, Ashburn, VA, USA

14.8.1 CLEANING REQUIREMENTS IN THE ELECTRONIC INDUSTRY The global ban of the CFCs has fundamentally changed cleaning in the electronic industry. Manufacturing processes were developed with the goal of avoiding cleaning. However, this objective has only be partially realized. Furthermore, there are still a lot of different areas in the electronic industry where cleaning of assemblies is highly recommended and necessary. Examples of such applications can be found in the aviation and space industries and in addition in the rapidly growing telecommunication industry. When investigating the production process of printed circuit assemblies, there are several important cleaning applications that have varying degrees of impact on the quality of the manufactured assemblies (see Figure 14.8.1).

Figure 14.8.1. Cleaning applications in the electronic production.

Cleaning applications in the SMT assembly line typically start with the cleaning of the bare boards (1). The preliminary substrates have to be cleaned in order to remove a variety of residues (particles, chemicals used in the manufacturing process of the bare boards). These residues may have a negative impact on the subsequent production steps. A detailed analysis1 of different kinds of contamination which might occur on the bare boards revealed

14.8 Electronic industry

895

that these residues might cause failures in production steps far beyond the initial printing process. Therefore it is important to assure clean surfaces in each production step. The cleaning applications (2-4) outlined in the SMT process line, shown in Figure 14.8.1, refer to the soldering process. Within this production step the components are placed on the bare board using solder paste that is printed on the solder pads through a stencil. The stencil can be cleaned continuously during the printing process by using a stencil printer underside wiping system. However, it has definitely to be cleaned after removing the stencil from the printer. The solder paste is removed from the apertures of the stencils in order to assure an accurate printing image (3).2,3 If the assemblies are to be mounted on both sides, the components are placed on one side by using an epoxy adhesive (SMT adhesive), in order to prevent them from being accidentally removed during the reflow process (3).3 The soldering process is carried out in a specially designed reflow oven. The cooling coils which contain condensed flux residue of such reflow ovens also have to be cleaned (5). In the case of wave soldering processes, the flux contaminated solder frames in addition must be cleaned. The contamination that must be removed after the soldering process is predominantly from flux residues. The removal of flux residues from the soldered assemblies (6) is generally the most critical application. Thus it is the important that the assemblies are cleaned. A subsequent coating process demands a very clean and residue-free surface to assure long term stability of the coating against environmental stresses such as humidity.1 Asked for the reasons for cleaning most of the process engineers give the following answers: 1) Stencils and other tools, such as squeegees, are cleaned in order to assure a reproductive and qualitative satisfying printing process.2 2) Contaminations from printed circuit assemblies are removed to achieve long-term reliability. 3) Cleaning is an important manufacturing step within the SMT process line if subsequent processes such as coating and bonding are required. In recent times an additional process step, which cleaning definitely is, was regarded to be time and cost inexpedient. Therefore a lot off optimization techniques have been applied to the SMT production process to try to avoid cleaning. However, considering new technologies in the electronic industry like fine pitch, flip chip and micro-BGA applications, the above statement is no longer true. The question if the removal of flux residues from printed circuit boards is necessary can only be answered by carrying out a detailed process analysis including the costs for cleaning and the increased reliability of the products. A large number of global players in the electronic industry prefer the advantages of better field reliability of their products. This aspect is even more important taking into consideration the outstanding competition in these industries. Whether a product has to be cleaned, and, especially, when cleaning is necessary in the SMT production depends on the following factors. • Process costs: cleaning is an additional process step that demands additional investment and resources. However, failures and downtime in other process steps can be diminished. • Process reliability: due to a larger process window, the manufacturing process as a whole is more stable. • Product reliability: the functionality and long term reliability of the products has to be assured.

896


After the decision has been made in order to choose the most suitable cleaning process, the following two questions have to be answered in advance: • Which contamination has to be removed? • What are the requirements regarding cleanliness of the surface, especially taking the subsequent process steps into account. The most important consideration in achieving an optimal cleaning result, is the interaction between the chemistry of the cleaning agent and the type of contamination. This is explicitly shown in the below presented lock and key principle. However, the cleaning result is not only depended on the nature of the chemistry used for cleaning. Other important factors in the cleaning process are the type of cleaning equipment used and the applied Figure 14.8.2. Factors influencing the cleaning process. process parameters such as temperature, time and method of application of the cleaning agent (see Figure 14.8.2). In order to optimize a cleaning process for the specific requirements, detailed information on the nature and composition of the kind of contamination that has to be removed is necessary. This tailoring of the chemistry is achieved by carrying out time consuming screenings matching typical contamination with suitable solvents. Another very important aspect is the material compatibility of the cleaning agent and the substrates that need to be cleaned. The cleaning chemistry should not attack the different materials of the components in any manner. Consequently, a number of long-term compatibility tests have to be carried out to assure this very important process requirement. Also, more environmental and worker safety issues are ever increasing when evaluating a new cleaning process. Cleaning processes approved by the industry have to pass even higher standards than are demanded from the government. They have to show, under the worst case scenario, a large process window and a large process margin of safety with respect to personal and environmental exposure. If all of the above mentioned process requirements such as • excellent cleaning result, • long-term functional reliability of the substrates after cleaning, • material compatibility between cleaning chemistry and the substrates, and • no drawbacks due to environmental and safety issues should be fulfilled, a lot of time and know-how has to be invested. 14.8.2 AVAILABLE ALTERNATIVES Due to the global ban of the CFCs, a large number of different cleaning processes4 have been developed. Each of them has advantages and disadvantages regarding the above mentioned process requirements. The available cleaning processes can be divided into the following main groups: 1) aqueous processes, 2) semi-aqueous processes, 3) water-free processes based on solvents, and


897

4) special cleaning processes. In an aqueous cleaning process,5 cleaning is achieved by using water or a water-based cleaning agent. In order to avoid drying spots, rinsing is carried out with deionized water. By soldering with water-soluble solder pastes only hot water is necessary for cleaning. However, the use of an alkaline water-based cleaning agent is effective to remove more critical flux residues with low or high pressure spray in Figure 14.8.3. PCB cleaning. cleaning equipment. In semi-aqueous processes,4 cleaning is carried out using a high boiling mixture of organic solvents. Rinsing is achieved by using deionized water. The organic solvents used for these cleaners are predominantly based on alkoxypropanols, terpenes, high-boiling alcohols, or mixtures of different chemistries. In applications based on solvents such as isopropanol or alkoxypropanols6 a rinse with deionized water is not imperative, since these chemistries do not leave any residues if used for rinsing. Especially when using an uniformly boiling solvent-based cleaning agent, vapor rinsing with solvent vapor is possible. All there mentioned cleaning process will be discussed more detailed in the following sections. Less widespread applications in the electronic industry include: • Cleaning using supercritical carbon dioxide (SCF).7 • Cleaning using plasma.7 SCF cleaning alternative is especially useful for extremely sensitive and complex assemblies. At temperatures above 31°C and pressures above 73 bar, carbon dioxide transforms into a supercritical phase. Supercritical carbon dioxide reveals an extremely low surface tension. Consequently, the wetting of small gaps and complex assembly architectures can be achieved. However, the contamination that should be removed has to be nonpolar, and the compounds should be resistant to high pressures. Due to these basic limitations and the high costs of this cleaning process, this process does not play an important part in the electronic industry. The advantage of plasma cleaning is in almost waste-free removal of contaminations. The contaminations are transformed into carbon dioxide and water. However, this cleaning process is only suitable for removal of the small amounts of residue from the substrate surface. Combinations with aqueous cleaning processes are feasible and common in the field. 14.8.2.1 Water based systems; advantages and disadvantages 14.8.2.1.1 Cleaning with DI - water Cleaning without using any kind of cleaning agent in the electronic assembly process is possible when water-based fluxes are used for the soldering process. Since these aqueous fluxes are highly activated, aggressive residues occur after the soldering process. These residues have to be removed immediately to avoid severe corrosion of solder joints. The use of water-based fluxes is very common in the US electronic market due to the main advantage of a large soldering process window for soldering. DI-water is used for this purpose since it is able to solubilize the polar residues. However, physical energy, such as high-spray-pres-

898


sure, is in most cases necessary to assure complete and residue-free removal of the contamination. However, since pure deionized water reveals a relatively high surface tension, the wetting ability is extremely low.8 Consequently, being confronted with the situation of constant increase of the packaging densities on PCBs, flux residues especially, between or under components, are very difficult to be removed.8 As a result this presently predominant cleaning process will be in decline in the future. 14.8.2.1.2 Cleaning with alkaline water-based media The principle of this kind of cleaning process is based on the presence of saponifiers in the cleaning fluid. These saponifiers are able to react with the non-soluble organic acids, that are the main ingredients of the flux residues. In other words, the non-soluble flux residues are transformed into water-soluble soaps due to a chemical reaction between the saponifier and the flux. This reaction is called saponification. For such a cleaning process an effective rinse with deionized water is imperative to minimize the level of ionic contamination.1,5 However, the presence of alkaline components in the aqueous cleaning fluid may cause the phenomenon of corrosion on different metals (Al, Cu, Sn) resulting in the visible dulling or discoloration of solder joints and pads.9,10 The chemical nature of the saponifiers is either organic or inorganic. Organic saponifiers are for example monoethanolamine or morpholine.5 These soft bases transform the insoluble organic carboxylic acids to soluble soaps acting, in addition, as a kind of buffer to stabilize the pH of the cleaning solution. However, the above mentioned organic amines have an associated odor, and more important, are classified as volatile organic compounds (VOCs). Inorganic saponifiers based on the system sodium carbonate/sodium hydrogen carbonate are also very attractive alternatives exhibiting sufficient cleaning ability at pH-values below 11. By using these chemistries the VOC and odor issue can be avoided.5 Besides organic or inorganic saponifiers a large scale of different anionic, cationic or nonionic surfactants are present in water-based alkaline cleaning solutions. The main reason for adding surfactants is the requirement for lower surface tension in order to increase the wetting of the substrates. However, using surfactants in cleaning median in high-pressure spray in air cleaning applications, the issue of foaming has to be taken into consideration. 14.8.2.1.3 Aqueous-based cleaning agents containing water soluble organic components There is a large number of products on the market that are mixtures of water and water-soluble organic solvents. These media are recommended for the removal of solder paste, SMT adhesives and flux residues from stencils, misprints and populated reflowed PCBs.11 The water-soluble organic solvents are predominantly natural long chain alcohols, glycol ether derivatives or furfuryl alcohol. The basic principle of the cleaning process using such fluids is based on the removal of lipophilic contamination such as oil, grease, flux or adhesive through the organic components of the mixtures. After the removal from the surface of the substrates, the lipophilic contamination precipitates out of the water-based cleaning agent and can be removed by using a filtration devices. In order to guarantee a sufficient cleaning performance, these mixtures contain organic components up to 50%. Consequently, these formulations tend to be classified as cleaning agents with high content of volatile organic compounds (VOCs).


899

14.8.2.1.4 Water-based cleaning agents based on MPC® Technology (MPC = Micro Phase Cleaning) Cleaning agents based on MPC Technology can be formulated either neutral or alkaline. The media combine the advantages of solvent and aqueous-based formulations revealing an extremely large process window.2,3 The principle of MPC systems is shown in Figure 14.8.4. The MPC media are based on formulations consisting of alkoxypropanols and water. Their phase behavior is dependent Figure 14.8.4. MPC® cleaning technology. on the temperature, agitation, and the dragged in contamination. The microphase, which is generated through temperature or media agitation, is responsible for the removal of the lipophilic components of the contamination. The ionics are removed by the aqueous phase.12 Since the ability to keep the organic contamination in the microphase is limited through the degree of loading, the lipophilic substances are expedited into the aqueous phase. Due to their lipophilic character, they precipitate out of the fluid and can be removed by a simple filtration device. As a result the loading of the cleaning fluid with contamination is minimal thereby minimizing the amount of organic components that finally pollute the subsequent rinsing stages.3 Alkaline pH, that is essential for the saponification process, can be realized through biochemical buffer systems, predominately aminoalkanoles. Due to the described cleaning principle, extremely long bath life times can be guaranteed with excellent cleaning results.2,3,12 The MPC® media are tailor made for all kinds of different cleaning equipments such as high and low pressure spray systems, spray under immersion, and ultrasonic applications.3 These cleaning agents based on MPC® Technology also contain low quantities of volatile organic compounds. However, due to this state of the art technology the percentage of VOCs are reduced to a minimal level. 14.8.2.1.5 Advantages and disadvantages of aqueous cleaning media Prior to the implementation of a cleaning process in a production line, the following criteria need to be taken into consideration: 1) Cleaning performance The substrates that have to be cleaned should be free of all kinds of residues that might have some negative influence on the functionality of the substrates. The contamination consist predominantly of flux residues, solder paste or SMT adhesive. Another important aspect with respect to the cleaning result is the compatibility between substrates and the cleaning chemistry. It is painless to remove the contamination while destroying the substrates. 2) Process reliability and process costs Every engineer tries to implement a process with a wide process window, and minimizing potential problems during production. While evaluating a cleaning process, the process parameters should be stable over a long period of time. As a result the equipment down time can be minimized, due to maintenance thereby optimizing the overall process cost.

900

Martin Hanek, Norbert Löw, Andreas Mühlbauert

3) Environmental aspects A cleaning agent should not only be considered based on its content of volatile organic compounds. A medium that does not contain any VOCs may on the one hand have no influence on air pollution. However, on the other hand, soil and water may be influenced. A VOC containing cleaning fluid has an inherently longer bath life. Consequently, the overall environmental calculation for some VOC containing cleaning fluid are better than by using VOC-free media. 4) Personal safety Another very important aspect for the evaluation of a new cleaning process is the issue on personal safety. The following topics have to be taken into consideration: • No risks during handling of the cleaning agent, especially regarding toxicity and flash point. • Low odor of the cleaning fluid. If the evaluated cleaning agent fulfils these important requirements, the chemistry will be accepted by the operator. Taking the discussed issues (1-4) into account, the advantages (+) and disadvantages (-) for the different aqueous cleaning fluids are as follows: a) Aqueous fluids using amines for saponification: + excellent removal of flux residues and unsoldered solder paste + predominantly VOC-free formulations low solid flux residues are difficult to remove limited removal of adhesives due to curing of the adhesive short bath life times resulting in high costs for media disposal b) Aqueous fluids using inorganic saponifiers and buffer systems: + satisfying cleaning results on flux residues and solder paste + VOC-free formulations + longer bath life due to constant pH level caused by the buffering systems low solid flux residues and SMT adhesives are difficult to remove c) Mixtures of organic solvents and water: small process window regarding flux residues and SMT adhesives VOC containing mixtures short bath life d) pH neutral and alkaline fluids based on MPC® Technology: + extremely large process window regarding flux residues, solder paste and adhesive removal + extremely long bath life-times due to previous described cleaning principle + economical cleaning process small amounts of VOCs. 14.8.2.2 Semi-aqueous cleaners based on halogen-free solvents, advantages and disadvantages A semi-aqueous cleaning process consists of a wash cycle using a mixture of different organic solvents followed by a rinse with deionized water. Organic contaminations, predominantly rosin flux residues, are removed through the lipophilic solvents, whereas the ionics are minimized by the polar rinse media. The different kinds of solvent formulations that are


901

available on the market reveal an extremely wide process window. Almost all different fluxes that occur in the electronic assembly process can be removed efficiently. In principle, the media that can be used for semi-aqueous cleaning processes can be classified into two main groups:4 • solvent-based mixtures that are not soluble in water, and • water-soluble cleaning fluids. The water solubility of the solvent-based cleaning agent is very important for the treatment of the subsequent water rinse. 14.8.2.2.1 Water insoluble cleaning fluids The first media designed for this application were predominantly based on terpenes13 and different mixtures of hydrocarbons and esters. In order to increase the rinse ability of the solvent-based mixtures with deionized water, different surfactants were added.14 The principle of using a water insoluble cleaning fluid was based on the idea to avoid contaminating the rinse sections through the cleaning fluid. The lipophilic contamination should be kept in the organic cleaning media that can be easily separated from the rinse water by a skimming device. However, experience in the field revealed that this kind of media have some critical drawbacks. Since the surface tension of the organic cleaning formulations are lower than the surface tension of deionized water (approx. 72 mN/m), the cleaning fluid stays on the surface and will not rinse off. Consequently, the cleaning result is not acceptable with regard to ionic contamination and surface resistivities.1 Acceptable cleaning results can only be realized through complicated multi-stage rinse processes. Another important disadvantage is the potential risk of re-contamination of the substrate when lifting the substrate out of the rinse section, since the top layer of the rinse section consists of polluted cleaning media. Consequently, effective agitation and skimming of the rinse water is imperative to avoid this phenomenon.4 14.8.2.2.2 Water-soluble, water-based cleaning agents The solvents used for this application are based on water soluble alcohols, alkoxypropanols and aminoalcohols.4,6 The chemical structure of these compounds combines a hydrophobic and hydrophilic groups. Due to this ambivalent structure it is possible to optimize the water solubility and lipophilic character for an optimal cleaning performance. Consequently, cleaning fluids based on this principle reveal the largest process window. All different kinds of flux residues, especially low rosin fluxes, are easily rinseable with water resulting in an residue-free cleaning process. The quality and long-term stability of these processes are proven through many applications in the field.8,15,16 Applications and processes will be discussed more detailed later on. 14.8.2.2.3 Comparison of the advantages (+) and disadvantages (-) of semiaqueous cleaning fluids In order to quantify the effectiveness of the two main groups of solvent based cleaning agents that can be used for semi-aqueous processes, the following key aspects has to be taken into consideration: • cleaning performance, • process reliability, • environmental aspects, and

902


• personal safety. a) Non water soluble cleaning agents: + wide process window regarding fluxes, solder pastes and adhesives low rosin flux residues are critical to remove possibility of re-contamination of the substrate in the rinsing stage cleaning fluids contain volatile organic compounds. b) Water soluble solvent based cleaning agents: + extremely wide process window regarding all kinds of fluxes, solder pastes and adhesives + residue-free removal of the cleaning fluid due to excellent rinse ability VOC containing formulations 14.8.2.3 Other solvent based cleaning systems Besides the already mentioned solvent based cleaners, there are some special solvents on the market. However, due to their physical and toxicological character, most of them are used for special applications. The advantages and disadvantages of the different products are summarized below. 14.8.2.3.1 Isopropanol

Figure. 14.8.5. Zestron FA/IPA.

Isopropanol is sometimes used for the cleaning of assemblies. More common, however, is the use of this solvent in stencil cleaning applications. Isopropanol is often also used as a rinsing media when cleaning with a solvent for hybrid or ceramic applications prior to bonding (see Figure 14.8.5). However, when using isopropanol for any kind of cleaning application there are some essential disadvantages that have to be taken into consideration: • Flammability due to low flash point, • Requires explosion-proof cleaning equipment, • Small process window with respect to the ability to flux residues, • Limited removal of adhesives due to curing of the adhesives, and • White residues formation with some flux residues. 14.8.2.3.2 Volatile siloxanes Due to their extremely low surface tensions combined with excellent wetting character, siloxanes were recommended for electronic cleaning applications. While siloxanes can be


903

used for the removal of non-polar contaminations, ionic contaminations cannot be readily removed. Consequently, for an effective cleaning process, formulations containing siloxanes and alkoxypropanols are needed. However, the use of siloxanes containing cleaning fluids does not play an important role in electronic assembly cleaning applications with respect to the above mentioned disadvantages. 14.8.2.3.3 Chlorinated solvents Chlorinated solvents such as trichloroethylene or 1.1.1-trichloroethane are nonflammable and excellent cleaning results can be achieved on especially non-polar residues. However, ionic contamination cannot be removed sufficiently. There are predominant environmental issues, in particular, their ozone depletion potential, which makes their use in electronic cleaning applications obsolete. 14.8.2.3.4 n-Propylbromide (nPB) Cleaning fluids based on formulations using n-propylbromide (nPB) have to be regarded very critically. Since these products also contain halogens, they cannot be considered to be real alternatives for chlorinated or fluorinated hydrocarbon mixtures. Due to these open environmental questions, cleaning applications using n-propylbromide formulations are still not very common, especially in Europe. 14.8.2.3.5 Alkoxypropanols Cleaning agents based on alkoxypropanols show extremely satisfying cleaning results due to their chemical structure. The lipophilic part of the molecule is capable of removing organic soils such as greases or oils from the substrate surface, whereas the hydrophilic part is able to remove the ionic contaminations. Furthermore, products based on this kind of chemistry show a large process window on all different types of flux residues. Different applications and processes will be discussed in detail in the following sections. 14.8.2.3.6 Hydrofluorinated ethers Hydroflourinated ethers are a new generation of cleaning fluids revealing no ozone depletion potential. Since they have no flash point there is no need to use explosion-proof cleaning equipment, thereby the cost of cleaning equipment is dramatically reduced. In addition, with their extremely low surface tension and viscosity, they are able to penetrate the tightest spaces of assemblies with high packing densities. Physical properties and different applications and processes will be discussed in more detail below. 14.8.2.3.7 Advantages and disadvantages of solvent-based cleaning agents Advantages: • Solvents exhibit a large process window on all different kinds of fluxes, adhesives and solder pastes • Most solvents can be regenerated through distillation, • There is no need of extensive treatment systems for the rinse water, Disadvantages: • Higher proven costs due to investment in explosion-proof cleaning equipment, • VOC containing solvents, Process: • Handling risks for the operator due to flammability, toxicity and odor, and

904


• Good material compatibility with respect to materials commonly found on assemblies. 14.8.3 CLEANING OF TOOLS AND AUXILIARIES The prohibition of CFCs in 1993 in Europe found the electronic industry scrambling to undertake appropriate actions. Various alternative technologies have asserted themselves during the conversion to environmentally friendly processes. Basically, two courses have evolved as alternative to the use of CFCs for changing applications: • New soldering techniques that avoid cleaning as a process have been developed and introduced in the electronic industry. • New environmentally friendly cleaning processes have been introduced. A large number of European companies decided in favor of no-clean processes, in spite of the associated disadvantages. In the USA cleaning has not be abandoned; it is still an important step in the SMT production line. However, with the increase of packing density and increased quality demands, more and more European companies are now returning to cleaning processes. In general, the purpose of changing is for the removal of contamination which might otherwise harm the operation of the electronic circuit assembly. Ensuring high reliability for a circuit assembly by effective cleaning is extremely important in such domains as the military, aviation, telecommunications, and in the automotive industry. The need for change to high reliability can be expected to grow appreciably because of different factors. These factors range from more complicated assemblies with higher density to the greatly accelerating use of conformal coatings in the automotive, niche computer, consumer electronics, and telecommunications markets. Dramatic growth is expected to occur in the USA, Europe and Asia. The applications where cleaning is necessary can be divided into three main applications: • the removal of flux residues from printed circuit assemblies (PCBs) after reflow or wave soldering; • the removal of solder paste and/or SMT-adhesive from stencils screens or misprints, and • the tool cleaning (dispensing nozzles, solder frames, cooling coils from reflow or wave soldering ovens, and squeegees). 14.8.3.1 Cleaning substrates and contamination In recent years stencil and misprint cleaning has become a significant issue for many users. Normally, in a typical PCB assembly line, the solder paste is printed through screens or stencils. As an alternative to glue dispensing, specially formulated SMT adhesives can now be printed through a stencil. Consequently, the stencils must be cleaned periodically to maintain the quality and yield of the process. However, the removal of SMT adhesive gives rise to completely new demands on the cleaning process and the condition of the stencils. The following specifies the points that must be observed in connection with stencil cleaning. Great importance has therefore been given to comparison of different cleaning chemistries, cleaning equipment, and comparison of manual and automated cleaning. Ever-increasing component density and the result expected from the printed image give rise to stringent requirements regarding stencil cleanliness. A flawless printed image is very important because misprinted boards can no longer be used without cleaning. This can be very expensive, particularly if components have already been mounted onto one side of a board. As a consequence, stencil printing for the application of solder paste on SMT boards


905

has become a standard. Stainless steel stencils with apertures that are either etched or cut with a laser are used for this purpose. They can also be used for fine-pitch apertures. Studies have revealed that more than 60% of the defects arising in the SMT process are caused by inadequate paste printing. This, in return, can be traced to insufficient cleaning that is often done manually. Consequently, a well working stencil cleaning process is an imperative. With regard to the cleaning process it is very important that no paste or cleaner residues remain in a stencil aperture, as this will inevitably cause misprints (see Figure 14.8.6, Figure 14.8.6. Left: Residues of SMT adhesive cured in the aperture of a stain- right). The printing of adless steel stencil. Right: Stainless steel stencil encrusted with solder paste. hesives, on the other hand, requires some points to be taken into consideration since this method has to produce different dot heights. The ratio between the stencil thickness and a hole opening is the decisive factor with this technology. Stencils for adhesive application are made of either stainless steel or plastic and are usually thicker than those used for paste printing. Cleaning is complicated by the thickness of the stencils (often exceeding 1 mm). Consequently, small holes require a cleaner with intense dissolving properties to ensure that the adhesives are completely removed (see Figure 14.8.6, left). Although cleaning of stencils improves the printing results drastically, the production of misprints during the printing process cannot be avoided completely. However, the cleaning of misprinted assemblies is an application that is still frequently ignored. This inFigure 14.8.7. Left: Misprinted PCB, Right: Via filled with solder paste. volves the removal of misprinted or smeared solder paste or SMT adhesive. The removal of solder paste as the most widespread application gives rise to the following demands: • No solder paste must remain on the surface, or the vias (see Figure 14.8.7). • The surface must be perfectly clean and dry to ensure immaculate renewed printing. • The solder flux on double-sided PCBs must also be removed from the side where soldering has been finished. 14.8.3.2 Compatibility of stencil and cleaning agent In some cases, repeated washing can damage stencils cleaned in automated washing stations. This damage usually occurs through degradation of the frame-to-screen glue. This degradation occurs at the junction of the stencil and the mesh and also at the junction of the mesh and the frame. The causes for this degradation lay in two factors. First, the high temperatures that are needed to clean stencils with aqueous cleaners cause thermal stresses

906


along the glue boundaries. Secondly, the glue itself, which in most cases is a cured epoxy material, may dissolve in the cleaning medium. The combination of these two factors can lead to severe stencil defects. A protective coating can be applied to the glue in cases where it is known that the glue is not stable towards the cleaning medium. Advanced cleaning agent manufacturers work closely with stencil manufacturers to avoid this problem. However, testing in advance of installed cleaning operation ensures that no unexpected stencil defects will affect the production process. 14.8.3.3 Different cleaning media When selecting a cleaning process, the chemistry should be selected first, because the cleaning chemistry has to be adopted to the chemical ingredients of the contamination. Only after the evaluation of the proper chemistry, can the method of applying the cleaning agent be evaluated. Solvent-based cleaning agents such as hydrocarbons, alcohols, terpenes, and esters offer properties that make them effective cleaning agents for the electronics industry. Low surface tension, high solubility, and ease of drying allow organic solvents to clean effectively where water-based cleaning agents are less effective. These benefits are very similar to the physical properties of the chlorofluorocarbons (CFC) that were once in widespread use in electronics manufacturing operations. Solvents, however, have several significant drawbacks that have limited their use including: • Flame and explosion hazards, • Higher equipment costs, due to required explosion protection, • Personal exposure concerns, • Material costs, and • Disposal of used solvents. And while water-based cleaning agents have been much improved in recent times, their performance does not always match that of their solvent forerunners. In fact, an ideal water-based cleaning agent would combine the physical properties of solvent-based cleaning agents with the safety and convenience of water-based materials. Solvent-based cleaning agents Cleaning with solvents such as isopropanol or acetone still remains the most widely applied method because any type of solder paste or SMT adhesive can be removed from the stencils by specially formulated mixtures. Consequently, the cleaning process window for such modern solvents is sufficiently wide to ensure the required results when changing over to another paste or adhesive. Normally, such cleaning processes are performed in machines designed specifically for solvent cleaning. The use of cleaners with flash points in excess of 104ºF (40°C) means that the machines can be set up without having to maintain an explosion protected area. Moreover installation in a special room may be necessary. However, the handling requirements of highly flammable cleaners, for such as isopropanol (IPA, flash point of 54ºF, 12°C), are far more critical, and cleaning is not so effective, particularly with fine-pitch stencils and SMT adhesives. Besides, modern solvent cleaners have been formulated in such a manner that health hazards are minimized - provided, they are properly used. Water-based cleaning agents The benefits of cleaning with water-based agents are straightforward. They include: • Ease of use, • Absence of fire or explosive hazards, • Simple disposal,


907

• Environmental and personal safety rules. In some cases even pure water can be used as the cleaning medium (presupposing that the contaminants are fully water-soluble). In most cases, however, water requires the addition of chemical or mechanical energy to obtain good cleaning performance. This can come in the form of: • A cosolvent, • A chemical activator, • Elevated temperature, • Mechanical action, or • High-pressure spray or ultrasonics. Contaminants such as oils, greases and SMT adhesives that feature high organic matter content are not readily soluble in water. Adding chemical activators such as saponifiers to the formulation can increase their solubility. Unfortunately, saponifiers require in most cases elevated temperatures (38-60°C) in order to react with the contamination. However, temperatures in this range will most likely damage the stencil. Saponifiers improve cleaning by reducing the surface tension of the cleaning solution, permitting the agent to penetrate void spaces and stand-off gaps that pure water cannot reach. The saponifiers in water-based cleaning agents are either organic or inorganic. Organic saponifiers are, in most cases, fully miscible with water and can easily be rinsed from the stencil. Any residues remaining after rinsing evaporate during drying. Inorganic saponifiers are usually water-soluble materials applied as a solution in water. They are nonvolatile, and residues remaining after rinsing might leave contamination on the substrate. The alkaline saponifier is consumed during the cleaning process and requires constant replenishment. In addition, the high pH value and the elevated application temperatures can also cause an oxidative attack of the aluminum frame of the stencil. Elevated temperatures are also commonly used to increase cleaning performance. An increase in temperature will lead to a corresponding increase in the solubilizing properties of the cleaning medium. There is of course a point at which the thermal stress induced by high temperature cleaning can damage the stencil adhesion. Lastly, high pressures or ultrasonics are used as a way to remove contaminants by transferring mechanical energy to particulate contamination. Pressure helps to force the cleaning agent into void spaces and stand-off gaps. Pressure can only marginally improve cleaning performance when the contaminant is not in particulate form or when solubility is the limiting factor. Water-based cleaners based on MPC® technology Satisfactory results in stencil cleaning with aqueous systems can only be achieved for the removal of solder paste. Most aqueous systems used to remove adhesives failed in the past because the water caused the adhesives to set. Micro Phase Cleaning (MPC)2,3,12 refers to the use of a cleaning agent formulated to undergo a phase change at elevated temperature. When heated above a threshold temperature (typically 100-120oF, 40-50°C) a microphase cleaner changes from a clear colorless solution to a turbid milky mixture. The phase transition produces a cleaning mixture that exhibits the properties of both solvent-based and water-based cleaners. “Water-like” properties allow the cleaning agent to effectively dissolve ionic contaminants and remove them from the substrate’s surface.

908


Non-polar and organic residues are removed by contact with the hydrophobic or “solvent-like” phase.12 No specialized equipment is needed to handle microphase cleaning agents. In most cases a microphase cleaning agent can be used as a drop-in replacement in existing equipment. Microphase cleaners are effective in all types of cleaning equipment capable of applying the cleaning agent in a liquid form. This includes spray and ultrasonic equipment. Table 14.8.1. Comparison of different cleaner types Cleaning Agent

Pros

Cons

Modern solvents

Removal of adhesives and solder paste Wide process window Cleaning at room temperature Do not oxidize/corrode stencil

Flammable Emits solvent vapors into work area Slow drying VOC (volatile organic compound) Ozone depletion potential Require explosion proof equipment

Aqueous alkaline cleaner

Cheap Non-flammable Non ozone depleting No VOCs Mild odor

Do not remove adhesives Short bath life times Narrow process window Water rinse necessary

Aqueous based cleaners of MPC Technology

Removal of adhesives and solder paste Wide process window Non-flammable Rapid drying Residue-free drying Long bath life

Agitation of the cleaner (spray, ultrasonic, spray under immersion, overflow) necessary

Table 14.8.2 presents the typical technical characteristics of different cleaning agents developed by Dr. O.K. Wack Chemie for stencil and misprint cleaning applications. 14.8.3.4 Comparison of manual cleaning vs. automated cleaning Printing screens are often cleaned manually (Figure 14.8.8) as users still shy away from the purchase of a cleaning machine because of the associated investment costs ($6,500 to $60,000). However, a more accurate assessment must take the following points into account: Table 14.8.2. Technical data of typical (1) The mechanical rubbing action of cleaning agents (water-based and solvent) the repeated manual cleaning impairs the stencil surface. This is particularly apparent ® ® Cleaning Agent Vigon SC200 Zestron SD300 with plastic stencils. These changes to the Chemistry water based solvent based stencil surface can result in misprints and shorten the service life of the stencil, makFlash point none 106°F ing it very difficult to track the source of the Appl. Temp. 77°F 77°F problem. (2) Precise repeatability of the cleanspray in air ultra- spray in air (ex Process sonic manual proofed) manual ing results is not guaranteed with manual


909

cleaning because manual applications are not based on precisely defined cleaning processes. This can also result in fluctuations of the printing result. (3) Manual cleaning must be completed with the utmost care. Such care costs time and money, thus making it very expensive. The investments for a cleaning machine can be recovered within a very short period, independent of the number of substrates that are to be cleaned. Example: A worker requires 1 hour per day for stencil cleaning. On the basis of $40.00/h and 220 workdays, a small machine ($8,700) can be a worthwhile investment within a year. Some users clean misprinted assemblies manually. This can prove to be very labor and cost intensive when larger quantities are involved. The cleaning results and their reproducibility will be rather inconsistent. Manual cleaning of PCBs with vias and blind holes with a brush or cloth can mean that solder beads are actually rubbed into the holes. Such holes can only be satisfactorily freed by ultrasonic or other mechanical means. It is imperative that the via are free of solder particles, otherwise the consequential damage can be very expensive. 14.8.3.5 Cleaning equipment for stencil cleaning applications As previously mentioned, it may well prove to be an economical proposition to invest in a specially designed cleaning machine if this can be justified by the number of substrates that have to be cleaned. In general, PCB assemblers use the following different types of cleaning equipment for stencil and misprint cleaning: • Spray in air, • Spray under immersion, and • Ultrasonic. Spray in air cleaning equipment Spray in-air machines consist of either a single or multiple reservoirs containing wash and rinse fluids (detergents, solvents, saponifiers, water, etc.). The stencils are placed in a separate chamber, and the wash/rinse fluids are then pumped from the reservoirs into the stencil chamber, and delivered onto the stencil via low pressure ( 2000 µg/m3). Federal permits: All major sources of regulated solvents are required to have federally enforceable operating permits (FOP)14,15 (also referred to as Title V permits). State permits: Most states require state permits for facilities that emit listed air pollutants.14,15 In some states federal permits and state permits are combined, while in other states facilities are required to have both a state or county (air district) permit and a federal permit. As part of annual emission inventory reporting requirements, many states already require reporting of HAP and VOC because of their state implementation plan (SIP). 14.10.2.2.2 Clean Water Act (CWA; 33 U.S. Code 1251 et seq.) The CWA is the major law protecting the “chemical, physical and biological integrity of the nation’s waters.” Under it, the U.S. EPA establishes water-quality criteria used to develop water quality standards, technology-based effluent limitation guidelines, and pretreatment standards and has established a national permit program [National Pollution Discharge Elimination System (NPDES) permits; 40 CFR 122] to regulate the discharge of pollutants. The states have responsibility to develop water-quality management programs. For extraction solvents vegetable oil extracting facilities are covered by basic discharge effluent limitations [direct discharges to receiving waters or indirect discharges to publicly owned treatment works (POTW)], and stormwater regulations.15 The amount of solvent in effluent discharges and in stormwater (for those covered) needs to be determined and possibly monitored as part of an NPDES permit and as part of the visual examination or testing of stormwater quality. 14.10.2.2.3 Resource Conservation and Recovery Act (RCRA; 42 U.S.Code 6901 et seq.) RCRA subtitle C (40 CFR 261) is a federal “cradle-to-grave” system to manage hazardous waste (including provisions for cleaning up releases and setting statutory and regulatory requirements). Subtitle D covers nonhazardous wastes. Materials or items are hazardous wastes if and when they are discarded or intended to be discarded. The act requires generators, transporters, and disposers to maintain written records of waste transfers, and requires the U.S. EPA to establish standards, procedures, and permit requirements for disposal. The act also requires states to have solid waste management plans, prohibits open dumping, and requires the EPA to establish criteria for sanitary landfills. EPA under RCRA also regulates underground storage tanks that store or have stored petroleum or hazardous substances. Hazardous wastes are either listed wastes (40 CFR 261.30-.33) or characteristic wastes (40 CFR 261.21-.24). The U.S. EPA defines four characteristics for hazardous waste: ignitability (40 CFR 260.21); corrosivity (40 CFR 260.22); reactivity (40 CFR 260.23); and toxicity (40 CFR 260.24). Any waste that exhibits one or more of these characteristics is classified as hazardous under RCRA. The ignitability definition includes a liquid that has a flash point less than 60oC (140oF); the EPA included ignitability to identify wastes that could cause fires during transport, storage, or disposal (e.g., used solvents). All of the solvents in Table 14.10.5 have flash points less than 60oC, so all could be a RCRA ignitability waste.

14.10 Food industry

933

14.10.2.2.4 Emergency Planning and Community Right-to-Know Act (EPCRA; 42 U.S. Code 11001 et seq.) Enacted as Title III of the 1986 Superfund Amendments and Reauthorization Act (“SARA”), the Act mandates the EPA to monitor and protect communities regarding releases of chemicals into the environment. It requires states to establish emergency planning districts with local committees to devise plans for preventing and responding to chemical spills and releases. [“Superfund” is the Comprehensive Environmental Response, Compensation and Liability Act (CERCLA) of 1980 that gives the U.S. EPA authority to force those responsible for hazardous waste sites or other releases of hazardous substances, pollutants, and contaminants to conduct cleanup or other effective response actions.] Section 304 (40 CFR 355.40): Facilities are subject to state and local reporting for accidental releases, in quantities equal to or greater than their reportable quantities (RQ), of extremely hazardous substances (EHS) or CERCLA hazardous substances (40 CFR 302, Table 302.4) under Section 304. n-Hexane, cyclohexane, acetone, and some of the other solvents discussed are CERCLA hazardous substances and have CERCLA RQ for spills (Table 14.10.5). Section 311, 312 (40 CFR 370.20-.21): Business must make MSDSs, for chemicals that are required to have an MSDS, available to state and local officials. Since all of the solvents discussed require MSDSs under the OSHA HCS, all are covered by these requirements. Section 313 (40 CFR 372), Toxic Release Inventory (TRI): Businesses are required to file annual reports with federal and state authorities of releases to air, water, and land above a certain threshold for chemicals on the TRI/Section 313 list (40 CFR 372.65) by July 1 each year for the previous year’s releases.21 TRI requirements are triggered if a facility is involved in manufacturing with 10 or more full-time employees, manufactures, processes, or otherwise uses with one or more listed substance(s) in a quantity above the statutory reporting threshold of 25,000 lbs./yr (manufactured or processed) or 10,000 lbs./yr (otherwise used). Beginning with the 1991 reporting year, such facilities also must report pollution prevention and recycling data for such chemicals pursuant to Section 6607 of the Pollution Prevention Act (42 U.S. Code 13106). n-Hexane was added to the TRI list in 1994 with reporting for 1995 emissions.19 The other solvents discussed are not on the TRI list. The EPA can add new chemicals to or delete chemicals from the TRI list as it deemed necessary and any person may petition the EPA to add chemicals or delete chemicals from the list. 14.10.2.2.5 Toxic Substances Control Act (TSCA; 15 U.S. Code 2601 et seq.) If a chemical’s manufacture, processing, distribution, use, or disposal would create unreasonable risks, the U.S. EPA, under the TSCA, can regulate it, ban it, or require additional testing. TSCA mandates the U.S. EPA to monitor and control the use of toxic substances by requiring the Agency to review the health and environmental effects of new chemicals [referred to as “Premanufacturing Notice” or “PMN”; Section 5(a)(1) of TSCA] and chemicals already in commerce. The U.S. EPA also has Significant New Use Rules (SNUR) under Section 5(a)(2) of TSCA which provides a way for the U.S. EPA to restrict uses of a chemical substance already in commerce that are proposed for new uses. All of the solvents discussed are already commercially available, so a PMN would not apply; some could be

934

Phillip J. Wakelyn, Peter J. Wan

subjected to SNUR (40 CFR 721, subpart A), since some are not presently being used as extraction solvents in large quantities. Under Section 4(a) of TSCA, the U.S. EPA can require testing of a chemical substance or mixture to develop data relevant for assessing the risks to health and the environment. Section 8(d) of TSCA requires that lists of health and safety studies conducted or initiated with respect to a substance or mixture be submitted to the U.S. EPA. All new toxicological data of the effects of a chemical not previously mentioned must be reported immediately if the data reasonably supports the conclusion that such substance or mixture presents a substantial risk of injury to health or the environment [Section 8(e) of TSCA]. Testing (Section 4 test rule) was required for several of the solvents earlier (e.g., commercial hexane for which new toxicological information was reported to the U.S. EPA since 1992),22 and any new toxicological information will have to be reported to the U.S. EPA under Section 8(e) and 8(d). 14.10.2.3 Food safety In the U.S. the use of a solvent to extract oil, that is a human food product or used in a food product, from oilseeds and biological materials falls under the rules and regulatory jurisdiction of the U.S. Food and Drug Administration (FDA), which regulates all aspects of food, including food ingredients and labeling in the U.S. In order to be legally used as an oilseed extraction solvent in the U.S., a substance must have been subject to an approval by the U.S. FDA or the U.S. Department of Agriculture (USDA) during 1938-1958 for this use (“prior sanction”); be generally recognized as safe (GRAS) for this use; or be used in accordance with food additive regulations promulgated by the U.S. FDA. Many prior sanctions and GRAS determinations are not codified in the U.S. FDA regulations. However, extracting solvents used in food manufacturing, such as n-hexane, have been labeled as a food additive, solvent, defoaming agent, component of a secondary food and color additives, minor constituent, or incidental additives (i.e., “additives that are present in a food at significant levels and do not have any technical or functional effect in that food”) depending on the application. Incidental additives can be “processing aids,” (i.e., “substances that are added to a food during processing but removed from the food before it is packaged”). Most food-processing substances, including solvents, can be regarded as “incidental additives” and thus are exempt from label declaration in the finished food product. Even if exempt from label declaration, all extraction solvents must be used in accordance with the U.S. FDA good manufacturing practices (GMP; 21 CFR 100). In the U.S., the Flavor and Extract Manufacturers’ Association (FEMA) has conducted a program since 1958 using a panel of expert pharmacologists and toxicologists to determine substances that are GRAS. This panel uses all available data, including experience based on common uses in food. This safety assessment program (“FEMA GRAS”) is widely accepted and considered an industry/government partnership with the U.S. FDA.23 A number of papers published in Food Technology since 196124,25 list the substances that the panel has determined to be GRAS and the average maximum levels in parts per million (ppm) at which each has been reported to be GRAS for different categories of food. The U.S. FDA has not incorporated these substances in their regulations but does recognize the findings of the Expert Panel of FEMA as GRAS substances. Since vegetable oil and other human food grade oils undergo deodorization (steam distillation) and other purification processes (i.e., refining and bleaching) as part of the manufacturing process prior to being used as a food product, they should not contain any of

14.10 Food industry

935

the extraction solvent, if proper manufacturing practices are followed. (see Section 14.10.3.3 Processing crude oil, for more details.) Refining removes free fatty acids and other non-oil compounds (e.g., phospholipids, color, and trace metals); bleaching with acid-activated bleaching earth or clay (e.g., bentonite), removes color-producing substances and residual soaps; and deodorization, the last major processing step in edible oils refining removes volatile compounds (undesirable ingredients occurring in natural oils and those that may be imparted by prior unit processes or even storage, many of which are associated with undesirable flavors and odors).26,27 Most commercial deodorizers operate at a temperature of 245-275oC (475-525oF) under a negative pressure of 2-10 mm Hg.26,27 It has been reported that no hexane residue remains in the finished oil after processing due to its high volatility.28 In addition, animal-feeding studies with expeller and solvent-extracted meals have not indicated any adverse health affects related to the extraction solvent.29 Hexane has been used since the 1940’s as an Figure 14.10.1a. Flow diagram of oilseed ex- oilseed-extraction solvent on the determination that traction process from seed to crude oil and it is GRAS and it may also be subject to a prior sancmeal. tion. However, like many other food-processing substances, there is no U.S. FDA regulation specifically listing hexane as GRAS or prior sanctioned. GRAS status may be determined by a company (“GRAS self-determination”), an industry, an independent professional scientific organization (e.g., FEMA GRAS), or the U.S. FDA. The Federal Food, Drug and Cosmetic Act (FFDCA; 21 U.S. Code 321 et seq.) does not provide for the U.S. FDA to approve all ingredients used in food, and the U.S. FDA explicitly recognizes that its published GRAS list is not meant to be a complete listing of all substances that are in fact GRAS food substances. Although there is no requirement to inform the U.S. FDA of a GRAS self-determination or to request FDA review or approval on the matter, the U.S. FDA has established a voluntary GRAS affirmation program under which such advice will be provided by the agency. Solvents that do not have prior sanction, a GRAS determination, or a tolerance set, probably should be evaluated for compliance under food safety requirements, if a facility is considering changing its extracting solvent or using a solvent for the extraction of the various biological materials for specialty markets. 14.10.3 THE SOLVENT EXTRACTION PROCESS Three types of processing systems are used to extract oil from oil-bearing materials: expeller pressing, prepress solvent extraction, and direct solvent extraction. Only prepress solvent extraction and direct solvent extraction, which remove the oil from the conditioned, prepared seed with an organic solvent, will be discussed here1,27 (see Figure 14.10.1). Oil-bearing materials have to be prepared for extraction to separate the crude oil from the

936


Figure 14.10.1b. Flow diagram of oilseed extraction process. Overview of extraction operation and identification of emission sources.

meal. Careful control of moisture and temperature during processing must be exercised to maintain the quality of the protein in the meal and to minimize the damage to the oil. Crude oils are refined by conditioning with phosphoric acid and treating with sodium hydroxide (alkali-refining) (see Figure 14.10.2). Refined oil is bleached with activated clay to remove color pigments. Bleached oils are then deodorized by steam distillation. The refined, bleached, and deodorized oil (RBD oil) is used to produce finished products, e.g., salad and cooking oils, shortening and margarine. Some of the finished products also require the oil to be hydrogenated, which changes the consistency of the oil, and increases stability to oxidation, which extends the shelf life of the finished products. Also some of the oils are winterized to remove the higher melting constituents, which can be used in confectionary products; the winterized oil is less likely to become cloudy in refrigerated storage. 14.10.3.1 Preparation for extraction Storage: For optimum extraction and quality of oil, the oil-bearing material should be stored so that it remains dry and at relatively low temperature. If it is wet, it should be processed as soon as possible after harvest. Oils in the presence of water can deteriorate rapidly, forming free fatty acids and causing greater refining loss. Seed cleaning: The first step in the commercial processing of oilseeds is “cleaning”, to remove foreign materials, such as sticks, stems, leaves, other seeds, sand, and dirt using dry screeners and a combination of screens and aspiration. Permanent electromagnets are also used for the removal of trash iron objects. Final cleaning of the seed usually is done at the extraction plant just prior to processing.

14.10 Food industry

937

Figure 14.10.2. Flow diagram of edible oil processing.

Dehulling: After cleaning, it may be necessary to remove the seed’s outer seedcoat (hull). The seedcoat contains little or no oil, so its inclusion makes the extraction less efficient. Also, the next processing step is grinding to reduce particle size, and any tough seedcoats would interfere with this process. If the hulls are not removed prior to extraction, they will reduce the total yield of oil by absorbing and retaining oil in the press cake. An acceptable level of hull removal must be determined, depending on the desired protein level of the final meal. Hulls are removed by aspirator and undehulled seeds are removed from the kernels by screening and returned to the huller. Some meats still adhere to the hulls, which are beaten, then screened again to obtain the meat. Grinding, rolling, or flaking: After dehulling, the meats are reduced in size, or “flaked,” to facilitate oil removal. Proper moisture content of the seeds is essential for flaking, and if the moisture level is too low, the seeds are “conditioned,” with water or steam, to raise the moisture to about 11%. For solvent extraction, flakes are commonly not less than 0.203-0.254 mm (0.008-0.010 inch), which can be solvent extracted efficiently with less than 1% residual oil. Thinner flakes tend to disintegrate during the solvent extraction process and reduce the miscella percolation rate. Cooking: Prior to extraction, the flakes are heated. The purpose of cooking the flakes is: (1) cell walls are broken down, allowing the oil to escape; (2) oil viscosity is reduced; (3) moisture content is controlled; (4) protein is coagulated; (5) enzymes are inactivated and microorganisms are killed; and (6) certain phosphatides are fixed in the cake, which helps to minimize subsequent refining losses. Flakes are cooked in stack cookers to over 87.8oC (190oF) in the upper kettle. Flakes with high phosphatide content may benefit from being cooked at slightly lower temperatures to avoid elevating refining losses. The temperature of the flakes is raised to 110-132.2oC (230-270oF) in the lower kettles. The seeds are cooked

938


for up to 120 min. Overcooking lowers the nutritional quality of the meal and can darken both the oil and meal. Poor-quality seeds with high levels of free fatty acids cannot be cooked for as long a period as high-quality seeds because of darkening. Darker oil requires additional refining to achieve a certain bleach color. Expanders: Sometimes low shear extruders called expanders are used. This equipment has the capability to process both low- and high-oil content materials. The meats are fed into an extruder after dehulling, flaking, and cooking and are heated as they are conveyed by a screw press through the extruder barrel. The meats are under considerable pressure and temperature when they reach the exit of the extruder. The change in pressure as the material leaves the extruder causes it to expand and the oil cells are ruptured, releasing the oil, which is rapidly reabsorbed. The expanded “collets” produced are then cooled and extracted with solvent. 14.10.3.2 Oil extraction Prepress solvent extraction: In this process the oil-bearing material are first mildly pressed mechanically by means of a continuous screw press operation to reduce the oil by half to two-thirds of its original level before solvent extraction to remove the remaining oil in the pre-pressed cake. Pressing follow by solvent extraction is more commonly used when high oil content materials (e.g., canola/rapeseed, flaxseed, corn germ) are processed. Direct solvent extraction: This process involves the use of a nonpolar solvent, usually hexane, to dissolve the oil without removing proteins and other non-oil soluble compounds. Solvent extraction yields about 11.5% more oil than does the screw press method, and less oil remains in the meal. The cooked flakes or collets (if expanders are used) are mixed with hexane in a batch or continuous operation. The hexane vapor pressure limits the practical operating temperature of the extraction and its contents to about 50-55oC. The resulting miscella (oil-solvent mixture) and the marc (solvent laiden collets) are heated to evaporate the solvent, which is collected and reused. The oil is freed from the miscella, by using a series of stills, stripping columns, and associated condensers. The hexane-free oil (i.e., crude oil) is cooled and filtered before leaving the solvent-extraction plant for storage or further treatment. This is the crude oil normally traded in the commodity market. Occasional overheating of the oil-solvent miscella will cause irreversible color changes in the oil. 14.10.3.3 Processing crude oil Most crude edible oils, obtained from oil-bearing materials, consist primarily of triglycerides (triacylglycerols). The triglycerides (approximately 95% of the crude oil) are the constituents recovered for use as neutral oil in the manufacture of finished products. The remaining nontriglyceride portion contains variable amounts of other lipophilic compounds, such as free fatty acids (FFA), nonfatty materials generally classified as “gums,” phospholipids (phosphatides), tocopherols, color pigments, trace metals, sterols, meal, oxidized materials, waxes, moisture, and dirt. Most of these minor lipid components are detrimental to finished product color, flavor, and smoking stability, and so must be removed from the neutral oil by a processing/purification process. The object of the processing/purification steps is to remove the objectionable impurities while minimizing possible damage to the neutral oil and tocopherols and loss of oil during such processing. Lecithin and cephalin are common phosphatides found in edible oils. Soybean, canola/rapeseed, corn, and cottonseed are the major oils that contain significant quantities of phosphatides. Alkaline treatment used for FFA reduction is also capable of removing most

14.10 Food industry

939

of the phosphatides from these crude oils. Tocopherols are important minor constituents of vegetable oils, which are natural antioxidant that retard the development of rancidity. Refining, bleaching, and deodorization are the steps that are necessary if the oil is to be used in food applications. Oil that has only gone through these three steps is called “RBD” oil. Figure 14.10.2 illustrates the processing pathways. Refining: Refining involves the removal of nonglyceride materials (phospholipids, color, and trace materials) and FFA. The goal is to produce a high-quality refined oil with the highest yield of purified triglycerides. Refining is by far the most important step in processing. An improperly refined oil will present problems in bleaching and deodorization and reduce quality. Some solvent-extracted crude oils, including soybean or canola/rapeseed, contain approximately 2-3% gums, which are mainly phosphatides and require degumming. The principal phosphatides are lecithin and cephalin. Gums can cause problems through higher then necessary refining losses, or by settling out in storage tanks. The degumming operation exploits the affinity of most phosphatides for water, converting them to hydrated gums that are insoluble in oil and readily separated by centrifugal action. Lecithin can be recovered and concentrated from the gums in a separate solvent extraction process, usually with acetone. Either water-degummed oil or crude oil can be treated with sodium hydroxide solution to saponify free fatty acids that are subsequently removed as soapstock by a primary refining centrifuge. Conventional alkali refining is by far the most widespread method of edible oil refining. The success of the alkali refining operation is the coordination of five prime factors: (1) use of the proper amount of reagent (sodium hydroxide), (2) proper mixing, (3) proper temperature control, (4) proper residual contact time, and (5) efficient separation. Oil is alkali-refined by the addition of sodium hydroxide solution at a level sufficient to neutralize the FFA content of the oil. An excess of sodium hydroxide is required to reduce the color of the refined oil and to ensure the completion of the saponification reaction and to remove other trace elements. The amount and strength of the sodium hydroxide solution needed to neutralize the FFA is dependent on the amount of both FFAs and phosphatides present in the crude oil. Water-soluble soaps are formed in the primary reaction between the sodium hydroxide and FFAs. The hydratable phosphatides react with the caustic forming oil-insoluble hydrates. The caustic used in alkali refining is normally diluted to about 8-14% NaOH, although higher concentrations are occasionally used to reduce color. The proper amount of NaOH solution added to the oil will produce an adequately refined oil with the minimum of triglyceride oil loss. The amount of NaOH solution (neutralizing dose plus excess) is determined by experience and adjusted according to laboratory results. After the NaOH solution is injected, it is mixed for 6-10 minutes to ensure thorough contact. The treated oil is then heated to assist in breaking of the emulsion prior to separation of the soapstock from oil in continuous centrifuges. Any soap remaining, after the primary soapstock separation, is removed through continuous hot water washings. In this step, water is added at 10-15% at a temperature sufficient to prevent emulsification, generally 82-90.5ºC (180-195ºF). The oil is again separated from the soapy phase in water wash separators and drier prior to bleaching. Bleaching: The oil is further purified by “bleaching”, which removes color bodies and trace metals as well as entrained soaps, and products of oxidation that are adsorbed onto the surface of bleaching agents or adsorbents. Types of adsorbents most commonly used in-

940


clude neutral clay, acid activated clay, and activated carbon. The choice of adsorbent will depend on a balance between activity of the adsorbent, oil retention loss, and adsorbent cost. The process is generally carried out via batch or continuous bleaching. Adsorbent is mixed with the refined oil creating a slurry that is agitated to enhance contact between the oil and the adsorbent. This is generally carried out under a vacuum at 90-95ºC (194-203ºF) for 15-30 minutes. Vacuum bleaching offers the advantages of an oil with improved oxidative and flavor stability. Finally the adsorbent is filtered from the oil using pressure leaf filters precoated with diatomaceous earth. Spent clay is steamed for efficient oil recovery. Deodorization: Deodorization, which removes the volatile compounds along with residual FFA, is a critical step in ensuring the purity of any vegetable oil and improves flavor, odor, color, and oxidative stability. Many of the volatile compounds removed are formed by the auto-oxidation of fat, which produces aldehydes, ketones, alcohols, and hydrocarbons that are associated with undesirable flavors and odors. The process also is effective in removing any remaining pesticide residues or metabolites that may be in the oil. Deodorization, which can be conducted as a batch operation in smaller plants or as a continuous or semicontinuous process by larger deodorizing facilities, consists of a steam distillation process in which the oil is heated to 230ºC (446ºF) under a vacuum of 2-10 mm Hg. Steam is sparged through the oil to carry away the volatiles and provide agitation. The odor and flavor compounds, which are more volatile than the triglycerides, are preferentially removed. After deodorization and during the cooling stage, 0.005-0.01% citric acid is generally added to chelate trace metals, which can promote oxidation. Deodorized oils preferably are stored in an inert atmosphere of nitrogen to prevent oxidation. Tocopherols and sterols are also partially removed in the deodorization process. Tocopherols can be recovered from the deodorizer distillate in a separate operation. 14.10.4 REVIEW OF SOLVENTS STUDIED FOR EXTRACTION EFFICIENCY Research on solvents for extraction has been carried out for more than 150 years and has intensified since the first patent was issued to Deiss of France in 1855.1,3,48 In the early effort of selecting an extraction solvent, the availability, operation safety, extraction efficiency, product quality and cost were the major concerns. In recent decades, toxicity, bio-renewability, environmental friendliness have been added to the solvent selection criteria. Among the solvents tested, a majority of the candidate solvents were excluded on the ground of toxicity and safety. Only a handful of solvents are used to any degree. These are acetone, alcohol, hexanes, heptane, and water.1,4-7,49 Water is used in rendering of fat from animal tissues and fish and in coconut processing,49 alcohol for spice and flavorants extraction,5,49 acetone for lecithin separation and purification.4 For commodity oils derived from vegetable sources, only hydrocarbon solvents have been used since 1930’s. Acetone was used by an Italian cottonseed oil mill during the 1970’s.4 Aqueous acetone and acetone-hexane-water azeotrope were studied by the scientists at the Southern Regional Research Center of Agricultural Research Service, USDA during the 1960’s and 1970’s.4 The effort was stopped due to the cost of retrofit required, the difficulties in managing the mixture of solvents with the presence of water and product quality concerns - a strong undesirable odor associated with the acetone extracted meals.4 Ethanol and isopropanol were studied in the 1980’s as a potential replacement of hexane for oil extraction. Both were proven technically feasible but economically unacceptable.5,6 n-Hexane is listed as a HAP under the CAA14,15 (See Section 14.10.2.2.1 CAA) and there are other regulatory requirements. As a way to meet envi-

14.10 Food industry

941

ronmental regulations, a short term option to commercial hexane appears to be hydrocarbons with significantly reduced n-hexane content. 14.10.4.1 Hydrocarbon solvents Extraction of oils has largely relied on mechanical or heat rendering process for centuries.50 Increased demand of productivity to separate oils from oilseeds has been the principal factor driving the changes of oilseed processing from the ancient hydraulic press to a continuous screw press or expeller in early 1900’s.51,52 This operation still left more than 4-5% residual oil in the pressed cake.51 More complete recovery of oil can only be effectively accomplished by solvent extraction.53,54 Solvent extraction of oils had an early beginning. Deiss in France received a patent to extract fat from bone and wool with carbon bisulfide in 1855.53 A year later, Deiss received additional patents covering the extraction of oil bearing seeds. Large scale solvent extraction already was established in Europe in 1870.55 The earliest extractors were unagitated single-unit batch extractors of small capacity and not very efficient.56,57 These extractors were gradually modified by the addition of agitation. They were organized in a battery of ten batch extractors which can be operated in a countercurrent principle. Extractors of this type operated in European plants during the last three decades of the 19th century.57 Further development in solvent extraction technology was relatively slow until early twentieth century. Solvent extraction spread from Europe to various parts of the world including the United States and South America.57 The first extraction plant in the United States was used to recover grease from garbage, bones, cracklings, and other packing-house wastes and to recover residual oil from castor pomace.57 Wesson58 reported his efforts applying solvent extraction to recover cottonseed oil from 1889 till the close of World War I. During the 1930’s solvent extraction was introduced in the United States for the recovery of oil from soybeans and the German equipment of the continuous type was used almost exclusively.57-61 Just prior to World War II the installation of continuous solvent extraction equipment was greatly accelerated and throughout the period of the War new plants were erected in an effort to keep pace with the constantly increasing production of soybeans. All of the later installations have been of American manufacture and in a number of cases of American design.57-61 Solvents used in the early effort to extract grease and oils were diverse. Besides carbon bisulfide used by Deiss,53 chlorinated hydrocarbons, benzene, and alcohols were all being tried. Extracting oil from corn and cottonseed with both aviation gasoline and petroleum distillate was performed in the United States in 1915 and 1917 respectively.62 The hydrocarbon paraffins became the preferred solvents for oilseed extraction during 1930’s through the process of elimination.63-71 Due to the prominent defects of early solvent extraction: dark crude oil, strong solvent odor in meal and high cost associated with solvent loss, low boiling hydrocarbons such as propane and butane were recommended as oil extraction media.72 The flammability of hydrocarbons also prompted much research in 1940’s using chlorinated hydrocarbons as the extraction solvents73,75 before its meal was found unsafe as feed.75-77 For the purpose of improved protein and oil quality75,78 and of processing safety and biorenewable solvents,75,79,80 both ethanol and isopropanol were investigated as the oil extracting solvents. While these alcohols offer various advantages in product quality and process safety and are renewable, they are still not economically feasible to replace hydrocarbons as oilseed extraction solvents.81 Hexane rich solvent became popular for the oilseed industry,54,57,82,83 because it is the most efficient solvent, extracts minimum non-oil

942


material and is easy to separate from the crude oil and marc.63 A thorough comparison of various hydrocarbon solvents for cottonseed oil extraction on a lab scale basis was reported by Ayers and Dooley in 1948.84 A more recent study by Wan et al.85 used a laboratory scale dynamic percolation extractor which operates at the conditions similar to those applied in the oil mill practice. Plant trials of isohexane and heptane solvents versus hexane in a 300 tons/day cottonseed factory revealed some interesting findings.7 14.10.4.1.1 Nomenclature, structure, composition and properties of hydrocarbons Petroleum and natural gas are the most abundant and affordable sources of hydrocarbon. Sometimes naphtha is used to describe the low boiling liquid petroleum and liquid products of natural gas with a boiling range from 15.6oC (60oF) to 221oC (430oF). This large group of compounds can be structurally classified as aliphatic and aromatic. Aliphatic hydrocarbons include saturated alkanes (paraffins), unsaturated alkenes (olefins) and alkynes (acetylenes), and cycloparaffins (naphthenes). Paraffins can be linear such as n-butane, n-pentane, and n-hexane, and branched such as isobutane, isopentane, isohexane, etc. Example of olefins is ethylene; of cycloparaffins, cyclopentane and cyclohexane; and of an aromatic, benzene.8,86,87 These compounds are derived from natural gas and petroleum oils which normally contain thousands of hydrocarbons with molecular weight ranging from methane to about 50,000 - 100,000 Daltons. Upon refining, the crude petroleum is divided into hydrocarbon gases (methane, ethane, propane and butane), light distillates (naphthas and refined oils), intermediate distillates (gas oil and absorber oil), heavy distillates (technical oils, paraffin wax and lubricating oils), residues (petroleum grease, residual fuel oil still wax, asphalts and coke), and refinery sludges (acid coke, sulphonic acid, heavy fuel oils and sulfuric acid).88 Historically, various fractions of petroleum naphthas, pentane and hexane from the light naphthas, aviation gasoline and benzol from the intermediate naphthas, and aromatic hydrocarbon, benzene, have been tested for oil extraction.62-71 14.10.4.1.2 Performance of selected hydrocarbon solvents Factors affecting extraction: There is little theoretical basis to be followed for the extraction of oilseeds.66,89-91 The study of the extraction of oilseeds is complicated by the fact that the total extractible material is variable in quantity and composition.66,89 Composition of the early extracted material is nearly pure triglycerides. As the extraction progresses, increasing amount of non-glyceride material will be extracted.66,89 IT is believed that the majority of the oil from oilseed flakes is easily and readily extracted.66,90 While the thickness of flakes affects extraction rate, the concentration of miscella below 20% does not greatly increase the amount of time to reduce the residual oil in flakes to 1%.89 Good91 summarized much of the early effort in soybean extraction: (1) The first oil extracted is superior in quality to the last small fraction; (2) While other solvents have been used in the past, hexane has become the primary solvent due to a combination of properties; (3) Flake thickness is the most important factor in achieving good extraction results; (4) Higher extractor temperatures up to nearly the boiling point, improve extraction results; (5) Moisture control is important throughout the extraction process; (6) Heat treatment affects the total extractibles; and (7) The soaking theory of extraction indicates that weak miscellas are very effective in helping to achieve good extraction results. Particle size which relates to the surface area available for extraction and is obviously one of the most important factors for extraction study. Coats and Wingard92 noticed that par-

14.10 Food industry

943

ticle size was more influential when the seed grit was being extracted. When oilseed flakes were being extracted, the flake thickness would be a more important factor instead of size of the flakes. Moisture content in oilseed can affect the extraction results.93-95 Optimum moisture content of cottonseed meats for extraction was first reported by Reuther et at.94 to be from 9 to 10%. Work by Arnold and Patel95 indicated 7 to 10% to be the optimum moisture for cottonseed flakes and very little variation in extraction rate for soybean with moisture content between 8 and 12%. Wingard and Phillips96 developed a mathematical model to describe the effect of temperature on extraction rate using a percolation extractor as follows:

(

)

log(time,min) = n log Temp., oF + log k

or

time = k (Temp.)

n

[14.10.1]

where time is defined as the number of minutes required to reach 1% residual oil in the oilseed flakes. For all practical purpose, they concluded that the time in minutes required to reduce the oilseed to 1% residual oil content on a dry basis varied inversely with the square of the extraction temperature in degrees Fahrenheit. Evaluation methods: Except for the pilot plant batch or counter-current extraction described by various labs,54,74,78 most of the solvent extraction evaluation work found in the literature was done in one or several of the lab scale devices. The percolation batch-extraction apparatus of the Soxhlet type has often been used to evaluate the rate of extraction of hydrocarbon solvents such as the one described by Bull and Hopper.89 Wingard and Shand97 described a percolation type of extractor and a co-current batch extractor and claimed to be useful to study the factors influencing equipment design and plant operation as well as fundamental studies contributing to a general understanding of extraction. Wan, et al. modified the design of percolation type extractor to closely simulate a single stage counter current miscella extraction conditions as practiced in the factory.85 Co-current batch extractor with numerous variations was also frequently applied for the extraction properties of selected solvents which were often operated at room temperature.75,97 Soxhlet extraction84,85 and Soxtec System HT6 (Perstorp Analytical, Herndon, VA) were also frequently used to evaluate solvents.98 Soxhlet extractor allows vaporized and condensed pure solvent to percolate through oilseed sample. The temperature of the condensed solvent is normally lower than its boiling point. Depending upon the cooling efficiency of the condenser and the room temperature, the temperature of the condensed solvent and the temperature of the extracting solvent in the extractor largely varied from lab to lab. This extraction temperature variability was minimized with the Soxtec method by refluxing the oilseed sample in the boiling solvent for 15 minutes followed by Soxhlet type of rinsing for 35 minutes. In theory the Soxtec method is more efficient and better reproduced. However, the Soxtec method only utilized a 3 g oilseed sample. The heterogeneity of an oilseed sample could be a significant source of variation. Flakes of oilseeds were most frequently used for the solvent extraction studies. Sometimes, ground oilseed kernels through a specified sieve size was used.98 Residual oil content in the extracted flakes after a certain specified extraction condition or oil content in miscella (mixture of oil and solvent) was examined and the percentage of total oil extracted was calculated.89-97 The total extractable oil of flakes was determined by four hours Soxhlet extraction. Wan et al.85 used a precision densitometer to determine the miscella concentration (percent of oil in miscella by weight) after a given time of extraction from which the per-

944


centage of oil extracted from cottonseed flakes was calculated. From these data, Wan et al.85 was also able to estimate the initial rate of extraction and final extraction capacity for each solvent as fresh and at selected initial miscella concentrations up to 30%. Bull and Hopper89 conducted extraction of soybean flakes in a stainless steel batch-extraction apparatus of the Soxhlet type with petroleum solvents, Skellysolve F (boiling range, 35 to 58oC) at 28oC and Skellysolve B (boiling range, 63 to 70oC) at 40oC. The extraction was carried out to permit the miscella obtained by each flooding of the flakes with solvent to be recovered separately. Their results showed that iodine number decreased and refractive index increased slightly with the extraction time which implied that more saturated fat was extracted during later stage of the extraction. Oils extracted during the later stages of the extraction were found to contain greater amounts of unsaponifiable matter and were rich in phosphatides, as high as 18% of the last fraction. Skellysolve B which is a hexane rich solvent demonstrated a much faster initial rate of extraction than that of Skellysolve F which is a pentane rich solvent and therefore, it took longer to complete the extraction for Skellysolve F. The fatty acid profile of each fraction showed a slight increase of saturated and a slight decrease of unsaturated fatty acid in the later fractions. Arnold and Choudhury82 reported results derived from a lab scale extraction of soybean and cottonseed flakes in a tubular percolation extractor at 135-140oF with pure, high purity and commercial hexane, and reagent grade benzene. They claimed that pure hexane extracted soybean slower than high purity and commercial hexane. During the first 60 minutes of extraction, benzene extracted more oil than the hexanes. However, at the end of 80 minutes, benzene extracted only slightly more than pure hexane but definitely less than the commercial hexanes. Similar results were obtained for the four solvents when cottonseed flakes were extracted. A laboratory extraction study of cottonseed flakes using various hydrocarbon solvents was reported by Ayers and Dooley.84 Soxhlet extractor and Waring blender were used for these experiments. Among the petroleum hydrocarbon solvents tested were branched, normal and cyclo-paraffins as well as aromatic hydrocarbons with various degrees of purity. They were: pure grade (99 mole percent purity) n-pentane, isopentane, cyclohexane, benzene, and n-heptane; technical grade (95 mole percent purity) neohexane, diisopropyl, 2-methylpentane, 3-methylpentane, n-hexane, and methylcyclopentane; technical grade (90 mole percent purity) cyclopentane; and commercial grade n-heptane, isohexanes, n-hexane, isoheptane and n-heptane. To assess the performance of these solvents, they used the following empirical formula: Quality-Efficiency Rating = 0.4 (Oil Yield Factor) + 0.4 (Refining Loss Factor) + 0.2 (Refined and Bleached Oil Color Factor)

[14.10.2]

When comparing the oil yield factor alone, 3-methylpentane was rated the best. When comparing the solvents based on the empirical Quality-Efficiency Rating formula, they concluded that methylpentanes (3- and 2-methylpentane) were superior extraction solvents for cottonseed oil. The normal paraffins, highly-branched isohexanes, cycloparaffins, and aromatics were progressively rated as less efficient than methylpentanes. Therefore, they recommended a tailor-made solvent for the extraction of cottonseed should exclude aromatic hydrocarbons, have low limits on cycloparaffin content, and consist largely of normal and isoparaffin hydrocarbons.

14.10 Food industry

945

A more recent study by Wan et al.85 using a laboratory scale dynamic percolation type of extractor (Figure 14.10.3) operated at the following conditions such as, temperature (5oC below the boiling point of each solvent) and miscella flow rate (9 gal/min/ft2), similar to those applied in the oil mill practice. Commercial grade hexane, heptane, isohexane, neohexane, cyclohexane, and cyclopentane were used to extract cottonseed flakes which had 5.8% moisture and 31.4% oil. When these solvents were tested near their boiling points, hexane apparently extracted cottonseed oil at a higher initial rate, > 94% oil extracted after 2 minutes, than all other solvents. Both heptane and hexane were able to extract more oil at the end of 10 minutes of extraction. Isohexane demonstrated to have adequate initial extraction Figure 14.10.3. Schematic of bench-scale dynamic percolation exrate (80% oil extracted after 2 tractor. minutes) and extraction capacity (93% oil extracted after 10 minutes of extraction) but is noticeably less effective than hexane. Similar to findings by Ayers and Dooley,84 results from the study by Wan et al.85 also demonstrates that neohexane, cyclohexane and cyclopentane performed distinctly less efficiently than hexane, heptane and isohexane. Conkerton et.al.98 tested commercial heptane versus hexane in a Soxtec extractor. Under this extraction condition, heptane actually extracted more oil than hexane from ground cottonseed kernel passed through a 20 mesh screen. The oil and meal quality were not appreciably affected by the higher temperature extraction of heptane. Plant scale results: Although hydrocarbon solvents have been used for oilseed extraction since the 1930’s, very little in plant operating data are available. During the spring of 1994, Wan et al.7 conducted plant trials with commercial heptane and isohexane at a 300 tons/day cottonseed crushing plant which routinely used hexane as the extraction solvent. Test results indicated that heptane performed well as an extraction solvent. However, it required extra energy and time to recover and consequently reduced the throughput rate of cottonseed being processed. Isohexane on the other hand was termed as an “easier” solvent by the plant engineers than hexane to operate. The plant also experienced a 40% steam savings and better than 20% throughput increase when it was operating with isohexane.7 This encouraging result prompted a second plant trial with commercial isohexane.45 The second plant trial was carried out at a cottonseed oil mill with a relatively new extraction and

946


miscella refining facility which was constructed in 1988 with a designed capacity of 500 tons/day but operated at only 270 tons/day due to limited delinting capacity. After week long testing with commercial isohexane, this plant experienced more than 20% natural gas usage and easily increased the throughput rate by close to 10% when compared with commercial hexane. This energy savings with commercial isohexane over commercial hexane may be largely attributed to the difference in the amount of water present in their corresponding azeotropes. Isohexane requires an additional step - isomerization to manufacture and will always be priced higher than hexane. But based on the two cottonseed oil mill trials, isohexane can be a cost efficient solvent.99 One additional benefit, the shorter residence time of the extracted cottonseed marc in the desolventizer/toaster because of the lower boiling range of isohexane will likely preserve more vegetable protein in the final meals which has been observed by both plants during the tests.7,45 The benefit in improved quality of oils were not obvious in both plant trials but might be realized with extended trials. Further evaluation of hydrocarbon solvents: As indicated in the study conducted by Wan et al.,85 the commercial cyclic hydrocarbons are the least effective extraction solvent than the branched and linear hydrocarbons. The comparison of extraction efficiency of pure isomers of hexane was conducted with the same single stage extractor as displayed in Figure 14.10.3. This was done to identify any unique structure-function characteristics of these pure components of commercial hexane and provide some guidance to the selection and tailored formulation for future commercial isohexane for the oilseed extraction industry. The extraction results for various pure isomers of six carbon paraffins using the single stage extractor indicated the following: (a) cyclohexane is noticeably less efficient in extracting cottonseed flake than all the other isomers; (b) slightly branched isomers, such as, 2-methyl and 3-methyl pentane, and methyl-cyclopentane are very slightly less efficient than n-hexane; and (c) highly branched isomers, 2,2-dimethyl and 2,3-dimethyl butane, are slightly less efficient than slightly branched isomers in extraction (Unpublished data). 14.10.5 FUTURE TRENDS In the future there most likely will be new demands for highly specialized extraction solvents as newly domesticated species that make useful novel oils30 and other products and new or altered biological products with enhanced nutritional and industrial properties will be developed through conventional breeding and genetic engineering for use as “functional foods”31 (e.g., phytosterols to achieve cholesterol lowering); as oils with altered lipid profiles32 (e.g., for lower saturated fat) or with more vitamin E; new drugs/nutraceuticals, industrial chemicals (e.g., fatty acids for lubricants, as cosmetics, coatings, detergents, surfactants, flavors, polymers, etc.); as sources for specialty chemicals; as value added products; etc.31-38 Genetically enhanced (GE) /biotech crops make up a growing share of the agricultural output.39 Biotechnology is the most powerful tool ever put in the hands of agricultural scientists. The ability to breed desirable traits or eliminate problematic ones can yield potentially spectacular benefits, such as various chemicals of importance including improved fats and oils, and vaccines and medicine, improved nutrition (e.g., in casaba, oilseeds, rice, sweet potatoes), and improved yields with the use of less agricultural chemicals. GE/biotech crops could be increasingly developed as biofactories for a wide range of products, including nutrients pharmaceuticals, and plastics. There is much promise for being able to produce products that would protect millions from disease, starvation, and death. However biotechnology and GM crops have become very controversial, and have run into serious

14.10 Food industry

947

problems in Europe, particularly in the UK.40 Europe,41 Japan, Korea, and Australia/New Zealand already have some restrictions and require some labeling. The U.S. is reviewing the issue.42 Thus even though this technology has great promise for increased use of new and existing solvents for extraction of products from diverse biological materials, there are also many potential problems because of misperceptions and misinformation. In the U.S. EPA is developing Maximum Available Control Technology (MACT) standards for vegetable oil processing that are likely to be finalized in 2001 (see Section 14.10.2.2.1, Hazardous Air Pollutants), with enforcement three years after promulgation.16,17 Commercial hexane, which is a HAP and a VOC, is presently the solvent used.16,17 To meet new and existing CAA requirements it is likely that extraction facilities will become much more efficient chemical engineering operations with upgraded equipment,43 more computerized monitoring and control for better quality management,44 and better environmental management/stewardship.19 In addition it is possible that alternate solvents (e.g., isohexane3,7,45) or lower n-hexane content commercial hexane (30-50% vs. 64%) will be used to meet these regulations. It is also possible that solvents like acetone, which is not a HAP or VOC and is not on the TRI list, will be more strongly investigated.16,17,46 In Europe trans fatty acid labeling of retail foods is required and in the U.S. FDA has proposed to label trans fats as saturated fat on the nutrition labeling panel required on packaged food sold at retail (64 FR 62,764; Nov. 17, 1999). This regulation if, promulgated as proposed, will result in reformulation of many products that could affect the vegetable oil producing and extracting industries. It is clear that the future has much uncertainty, while at the same time it offers much promise. It appears that there will be many potential changes that will put new demands on extraction solvents. Solvents that are more environmentally friendly, are nontoxic to plants, workers, and consumers, have specialized properties, have high solvent power at low temperatures (are easy to desolventize), etc., may have wider potential use in conventional extraction as well as specialized niche markets. REFERENCES 1 2 3 4 5 6 7 8

9

10

11 12

Technology and Solvents for Extracting Oilseeds and Non-petroleum Oils, P.J. Wan and P.J. Wakelyn, Eds., AOCS Press, Champagne, IL, 1997. M.A. Williams and R.J. Hron, Bailey’s Industrial Oils and Fat Products, 5th edn., Vol. 4: Edible Oil and Fat Products: Processing Technology, Y.H Hui,., Ed., John Wiley and Sons, Inc., 1996, p. 119. P.J. Wan, Hydrocarbon Solvents, in Technology and Solvents for Extracting Oilseeds and Non-petroleum Oils, P.J. Wan and P.J. Wakelyn, Eds., AOCS Press, Champaign, IL, 1997, p.170-185. R.J. Hron, Acetone, in Ibid, p.186-191. R.J. Hron, Ethanol, in Ibid, p.192-197. E.W. Lucas and E. Hernandez, Isopropyl Alcohol, in Ibid, p. 199-266. P.J. Wan, R.J. Hron, M.K. Dowd, M.S. Kuk, and E.J. Conkerton, J. Am. Oil Chem. Soc., 72, 661 (1995). Occupational Health and Safety Administration Field Operations Manual, Chapter IV: Violations, C. Health Standards Violations, (OSHA Instruction 2.45B CH-4, Dec.13, 1993), The Bureau of National Affairs, Washington, DC, 1994, pp. 77:2513-18. Occupational Health and Safety Administration Technical Manual, Section I- Sampling, Measurement Methods, and Instruments, Chapter 1 - Personal Sampling for Air Contaminants, Appendix I:1-6. Sampling and Analytical Errors (SAEs) (Issued by OSHA Instruction TED 1.15, September 22, 1995; amended by OSHA Instruction TED 1.15 CH-1, May 24, 1996). 1997 TLVs and BEIs, Threshold Limit Values for Chemical Substances and Physical Agents and Biological Exposure Indices, The American Conference of Governmental Industrial Hygienists, Cincinnati, OH, 1997, pp. 12-40. J.B Galvin, C.J. Kirwin, D.W. Kelly, INFORM, 6(8), 951 (1995). H.H. Schaumberg and P.S. Spencer, Brain, 99, 183 (1976).

948

13

14 15 16 17 18

19

20 21

22 23 24 25 26 27 28 29

30 31 32 33 34

35 36 37 38 39 40 41 42 43 44 45 46 47 48 49


J.B. Galvin, Toxicity Data for Extraction Solvents Other Than Isohexane/Hexane Isomers, in Technology and Solvents for Extracting Oilseeds and Nonpetroleum oils, P.J.Wan and P.J.Wakelyn, Eds., AOCS Press, Champaign, IL, 1997, p. 75-85. P.J. Wakelyn, Cotton Gin and Oil Mills Press, 92(17), 12 (1991). P.J. Wakelyn and L.A. Forster, Jr., Oil Mill Gaz., 99(12), 21 (1994). P.J. Wan and P.J. Wakelyn, INFORM, 9(12),1155 (1998). P.J. Wan and P.J. Wakelyn, Oil Mill Gaz., 104(6), 15 (1998). U.S. EPA, Study of Volatile Organic Compound Emissions from Consumer and Commercial Products, U.S. Environmental Protection Agency, Office of Quality Planning and Standards, Research Triangle Park, NC 27711, EPA - 453/R-94-066-A, March 1995. P.J. Wakelyn and P.K. Adair, Assessment of Risk and Environmental Management, in Emerging Technologies, Current Practices, Quality Control, Technology Transfer, and Environmental Issues, Vol. 1, Proc. of the World Conference on Oilseed and Edible Oil Processing, S.S. Koseoglu, K.C. Rhie, and R.F. Wilson, Eds. AOCS Press, Champaign, IL, 1998, pp. 305-312 n-Hexane, U.S. EPA Integrated Risk Information System (IRIS) Substance File, U.S. EPA, 1999 (www.epa.gov/ngispgm3/IRIS/subst/0486.htm). U.S. EPA, Toxic Chemical Release Inventory Reporting Form R and Instructions (Revised 1995 Version), U.S. EPA, Office of Pollution Prevention and Toxics, Washington, DC, EPA 745-K-96-001, March 1996, Table II, p. II-1 J.K. Dunnick, Toxicity Studies of n-Hexane in F344/N Rats and B6C3F1 Mice, National Toxicology Programs, U.S. Dept. of Health and Human Services, NTP TOX 2, NIH Publication No. 91-3121, 1991. A. H. Allen, GRAS Self-Determination: Staying Out of the Regulatory Soup, Food Product Design, (April 1996 supplement to Food Product Design, 6 pages) (1996). B.L. Oser and R.A. Ford, Food Technol., 27(1), 64 (1973). R.L. Hall and B.L. Oser, Food Technol., 19, 151 (1965). A.M. Galvin, in Introduction to Fats and Oils Technology, P.J. Wan, Ed., AOCS Press, Champagne, IL, 1991,pp. 137-164. L.A. Jones and C.C. King, in Bailey’s Industrial Oil and Fats Products, 5th edn., Vol. 2 Edible Oil and Fat Products: Oils and Oil Seeds, Y.H. Hui, Ed., John Wiley and Sons, Inc., 1996, pp. 177-181. H.W. Lawson, Standards for Fat and Oils, The AVI Publishing Co., Inc., Westport, CT, 1985, p. 34. S.W. Kuhlmann, M.C. Calhoun, J.E. Huston and B.C. Baldwin, Jr., Total (+)- and (-)- Gossypol in Plasma and Liver of Lambs Fed Cottonseed Meal Processed by Three Methods, J. Anim. Sci., 72 (suppl. 1), 145 (1994). D.J. Murphy, INFORM, 11(1), 112 (2000) M.A. Ryan, Today’s Chemist at Work, 8(9), 59 (1999). B.F. Haumann, INFORM, 8(10), 1001 (1997). B. Flickinger and E. Hines, Food Quality, 6(7), 18 (1999). C.T. Hou, Value Added Products from Oils and Fats through Bio-processes, Int. Symp. On New Approaches to Functional Cereals and Oils, Beijing, China, Nov. 9-14, 1997, Chinese Cereals and Oils Association, Beijing, China, 1997, p. 669. D.J. Kyle, New Specialty Oils: Development of a DHA-rich Nutraceutical Product, Ibid, p. 681. J.K. Daum, Modified Fatty Acid Profiles in Canadian Oilseeds, Ibid, pp. 659-668. C.M. Henry, Chemical Eng. News, 77(48), 42 (1999). R. Ohlson, INFORM, 10(7), 722 (1999). G.J. Persley and J.N. Siedow, Applications of Biotechnology to Crops: Benefits and Risks, CAST Issue Paper No. 12, Council for Agricultural Science and Technology, December 1999. M. Heylin, Chem. Eng. News, 77(49), 73 (1999). Anom., Official J. European Communities, 43(L6), 13 (2000). B. Hileman, Chem. Eng. News, 77(50), 31 (1999). P. Delamater, Oil Mill Gaz., 104(11), 34 (1999). P.J. Wakelyn, P.K. Adair, and S.R. Gregory, Oil Mill Gaz., 103(6), 23 (1997). M. Horsman, Oil Mill Gaz., 105(8), 20 (2000). R.J. Hron, P.J. Wan, and P.J. Wakelyn, INFORM, In Press, (2000). J.A. Dean, Dean’s Handbook of Chemistry, Thirteenth Edition, McGraw-Hill, Inc., New York, NY, 1985. R.D. Hagenmaier, Aqueous Processing, in Technology and Solvents for Extracting Oilseeds and Non-petroleum Oils, P.J. Wan and P.J. Wakelyn, Eds., AOCS Press, Champaign, IL, 1997, p.311-322. L. A. Johnson, Theoretical, Comparative, and Historical Analyses of Alternative Technologies for Oilseed Extraction, Ibid, p.4-47.

14.10 Food industry

50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99

949

D.K. Bredeson, J. Am. Oil Chem. Soc., 60, 163A (1983). Cottonseed and Cottonseed Products, A.E. Bailey, Ed., Interscience Publishers Inc., New York, pp. 5, 615-643,(1948). V.D. Anderson, U.S. Patent 647,354 (1900). Anonymous. J. Am. Oil Chem. Soc., 54, 202A (1977). J. Pominski, L.J. Molaison, A.J. Crovetto, R.D. Westbrook, E.L.D’Aquin, and W.F. Guilbeau, Oil Mill Gaz., 51(12), 33 (1947). O.K. Hildebrandt, Fette Seifen Anstrichm, 46, 350 (1939). M. Bonotto, Oil & Soap, 14, 310 (1937). K.S. Markley, Oil Mill Gaz., 51(7), 27 (1947). D. Wesson, Oil & Soap, 10, 151 (1933). E. Bernardini, J. Am. Oil Chem. Soc., 53, 275 (1976). K.W. Becker, ibid., 55, 754 (1978). K.W. Becker, Oil Mill Gaz., 84, 20 (1980). W.E. Meyerweissflog, Oil & Soap, 14, 10 (1937). W.H. Goss, ibid, 23, 348 (1946). W.H. Goss, J. Am. Oil Chem. Soc., 29, 253 (1952). R.P. Hutchins, ibid, 54, 202A (1977). G. Karnofsky, ibid, 26, 564 (1949). A.E. MacGee, Oil and Soap, 14, 322 (1937). A.E. MacGee, ibid, 14, 324 (1937). A.E. MacGee, J. Am. Oil Chem. Soc., 26, 176 (1949). A.E. MacGee, Oil Mill Gaz., 52, 17,35 (1947). A.E. MacGee, ibid, 67, 22 (1963). H. Rosenthal, and H.P. Trevithick, Oil and Soap, 11, 133 (1934). I.J. Duncan, J. Am. Oil Chem. Soc., 25, 277 (1948). O.R. Sweeney, and L.K. Arnold, ibid, 26, 697 (1949). A.C. Beckel, P.A., Belter, and A.K. Smith, ibid, 25, 7 (1948). L.L. McKinney, F.B. Weakley, R.E. Campbell, A.C. Eldridge, J.C. Cowan, J.C. Picken, and N.L. Jacobson, ibid, 34, 461 (1957). T.A. Seto, M.O. Shutze, V. Perman, F.W. Bates, and J.M. Saulter, Agric. Food Chem., 6, 49 (1958). F.K. Rao, and L.K. Arnold, J. Am. Oil Chem. Soc., 35, 277 (1958). E. W. Lusas, L.R. Watkins and K.C. Rhee, in Edible Fats and Oils Processing: Basic Principles and Modern Practices, D. R. Erickson, Ed., AOCS Press, IL, p. 56, 1990. R.J. Hron, Sr., S. P. Koltun and A. V. Graci, J. Am. Oil Chem. Soc., 59(9), 674A (1982). R.J. Hron, Sr., M.S. Kuk, G. Abraham and P. J. Wan, ibid, 71(4), 417 (1994). L.K. Arnold and R.B.R. Choudhury, ibid, 37, 458 (1960). K.S. Olson, Oil Mill Gaz., 85, 20 (1980). A.L. Ayers and J.J. Dooley, J. Am. Oil Chem. Soc., 25, 372 (1948). P. J. Wan, D. R. Pakarinen, R. J. Hron, Sr., and E. J. Conkerton, ibid, 72(6), 653 (1995). M.P. Doss, Physical Constants of the Principal Hydrocarbons, The Texas Company, Third Edition, New York. (1942). L.A. Johnson and E. W. Lusas, J. Am. Oil Chem. Soc., 60(2), 229 (1983). McGraw-Hill Encyclopedia of Science and Technology, 10, 71 (1960). W.C. Bull and T.H. Hopper, Oil and Soap, 18, 219 (1941). H.B. Coats and G. Karnofsky, J. Am. Oil Chem. Soc., 27, 51 (1950). R.D. Good, Oil Mill Gaz., 75, 14 (1970). H.B. Coats and M.R. Wingard, J. Am. Oil Chem. Soc., 27, 93 (1950). W.C. Bull, Oil and Soap, 20, 94 (1943). C.G. Reuther, Jr., R.D. Westbrook, W.H. Hoffman, H.L.E. Vix and E.A. Gastrock, J. Am. Oil Chem. Soc., 28, 146 (1951). L.K. Arnold and D.J. Patel, ibid, 30, 216 (1953). M.R. Wingard and R.C. Phillips, ibid, 28, 149 (1951). M.R. Wingard and W.C. Shand, ibid, 26, 422 (1949). E.J. Conkerton, P.J. Wan and O.A. Richard, ibid, 72, 963 (1995). P.J. Wan, INFORM, 7, 624 (1996).

950

George Wypych

14.11 GROUND TRANSPORTATION George Wypych ChemTec Laboratories, Inc., Toronto, Canada

The ground transportation industry in the USA is dominated by truck freight (78.6%). Other methods of transportation include: rail (7.9%), water (5.2%), air (4%), pipeline (2.2%), and other (2.1%). Solvents are used and solvent wastes and emissions are generated during refurbishing and maintenance. Rail car refurbishing involves stripping and painting. Paint is usually removed by mechanical means (steel grit blast system) but solvents are occasionally used. Solid wastes are generated from latex paint wastes but hazardous wastes are also generated from solvent-based paints and thinners. Parts cleaning is mostly done using mineral spirits. Waste solvents are sent off-site for reclamation. Truck maintenance work usually requires a parts washer which may involve either a heated or ambient temperature solvent, hot tank, or a spray washer. In the solvent tank washer, solvent (usually mineral spirits, petroleum distillates, and naphtha) is recirculated from solvent tank. Spent solvent is usually replaced monthly. Carburetor cleaning compounds contain dichloromethane. Tanker cleaning often involves a solvent spray. The ground transportation industry employs a large number of people (more than 2 million in the USA). It is one of the less polluting industries. It generates 0.3% of VOC released by all major industries combined (about half of that of the aerospace industry). Most solvents used are of low toxicity. Good system of collection and reclamation of solvent wastes is done effectively and this is the major reason for the relatively good performance of the industry.

REFERENCES 1 2

EPA Office of Compliance Sector Notebook Project. Profile of the Ground Transportation Industry. Trucking, Railroad, and Pipeline. US Environmental Protection Agency, 1997. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.

14.12 INORGANIC CHEMICAL INDUSTRY George Wypych ChemTec Laboratories, Inc., Toronto, Canada

This industry has two major sectors: inorganic chemicals and chlor-alkali. Inorganic chemicals are often of mineral origin processed to basic chemicals such as acids, alkalies, salts, oxidizing agents, halogens, etc. The chlor-alkali sector manufactures chlorine, caustic soda, soda ash, sodium bicarbonate, potassium hydroxide and potassium carbonate. The major processes in this industry do not use solvents but there are many specialized auxiliary processes which use solvents. Tables 14.12.1 and 14.12.2 give information on the reported solvent releases and transfers from inorganic chemical industry.

14.13 Iron and steel industry

951

The tables show that the industry, which operates almost 1,500 plants and employs over 110,000 people, has minimal impact on global emission of VOCs. Consequently, the industry does not have any major initiative to deal with solvent emissions or wastes. Future safety improvements concentrate on non-solvent issues.

REFERENCES 1 2

EPA Office of Compliance Sector Notebook Project. Profile of the Inorganic Chemical Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.

Table 14.12.1. Reported solvent releases from the inorganic chemical plants in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

33,000

hexane

2,000

carbon tetrachloride

3,000

methanol

574,000

chloromethane

2,600

methyl ethyl ketone

460

dichloromethane

12,500

N-methyl-2-pyrrolidone

180

ethyl benzene

110

toluene

12,000

ethylene glycol

1,800

xylene

1,500

Table 14.12.2. Reported solvent transfers from the inorganic chemical plants in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

780

methyl ethyl ketone

9,000


6,400


8,700

dichloromethane

5,000

toluene

6,000

ethylene glycol

12,000

xylene

96,000

14.13 IRON AND STEEL INDUSTRY George Wypych ChemTec Laboratories, Inc., Toronto, Canada

With almost 1,400 plants, the US iron and steel industry is very diverse industry having total sales of $100 billion and over 400,000 employees. Figure 14.13.1 is a schematic diagram of the iron and steel making process. Only one stage − finishing − employs solvents. The finishing stage includes processes to remove mill scale, rust, oxides, oil, grease and soil prior

952

George Wypych

Figure 14.13.1. Schematic diagram of operations in the iron and steel manufacturing process. [Reproduced from EPA Office of Compliance Sector Notebook Project. Reference 1.]

Table 14.13.1. Reported solvent releases from the iron and steel plants in 1995 [Data from Ref. 2] Solvent benzene

Amount, kg/year 321,000

Solvent N-methyl-2-pyrrolidone

Amount, kg/year 3,600

n-butyl alcohol

26,000

polycyclic aromatic compounds

2,400

cresol

1,800

tetrachloroethylene

91,000

dichloromethane

318,000


17,000

ethylbenzene

5,000

trichloroethylene

620,000

methanol

241,000

toluene

261,000

methyl ethyl ketone

358,000

xylene

168,000

to coating. Methods used include solvent cleaning, pressurized water or air blasting, cleaning with abrasives, and alkaline or acid pickling. Tables 14.13.1 and 14.13.2 give information on the reported solvent releases and transfers from the iron and steel industries. Not all the solvents listed in the tables are used in processing. Some are by-products of coke manufacture from coal. Benzene and polycyclic aromatics compounds are by-products. Strong solvents such as methyl ethyl ketone, toluene, xylene, and trichloroethylene are typical of those used in cleaning processes. There is no program formulated by the industry to reduce amounts of solvents used.

REFERENCES 1 2

EPA Office of Compliance Sector Notebook Project. Profile of the Iron and Steel Industry. Trucking, Railroad, and Pipeline. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.

14.14 Lumber and wood products

953

Table 14.13.2. Reported solvent transfers from the iron and steel plants in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

3,000


11,500

n-butyl alcohol

1,400

polycyclic aromatic compounds

3,820,000

cresol

12

tetrachloroethylene

20,000

dichloromethane

14,500


3,600

ethylene glycol

197,000

trichloroethylene

165,000

ethylbenzene

550

toluene

11,500

methanol

25

xylene

14,000

methyl ethyl ketone

66,000

14.14 LUMBER AND WOOD PRODUCTS - WOOD PRESERVATION TREATMENT: SIGNIFICANCE OF SOLVENTS Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg Institut fòr Allgemeine Hygiene und Umwelthygiene Universit¬t Tòbingen, Tòbingen, Germany

Gerhard Volland Otto-Graf-Institut, Universit¬t Stuttgart, Stuttgart, Germany

14.14.1 GENERAL ASPECTS Wood preservation is based on various fundamental principles, e.g., construction aspects such as exposure to humidity, selection of different types of wood products according to their durability, and chemistry of wood preservatives. Important groups of chemical wood preservatives are water-soluble and solvent-based substances.1 The main requirements of chemical wood preservatives are:1 • Stability, especially chemical stability. • Resistance to environmental conditions, e.g., light or heat. • Penetration into the wood products. • Effectiveness against wood attacking agents (e.g., insects, fungi, bacteria). • Compatibility with other construction components, e.g., paints, adhesives, and fasteners. • Construction aspects, e.g., corrosion. • Minimal environmental impact, e.g., minimum emissions or minimum environmental pollution. • Ability to work with a range of wood products. • Case applications, e.g. fundamental differences of indoor and outdoor coatings. • Having favorable visual aspects, e.g., surface properties, color, uniformity, influence on grain pattern, etc. All requirements cannot be fulfilled completely by the various wood preservatives. Therefore wood preservatives should be selected according to the particular case.

954

Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg, Gerhard Volland

14.14.2 ROLE OF SOLVENTS 14.14.2.1 Occurrence Various solvents are added to wood preservatives. Only a limited number of wood preservatives are authorized by governmental agencies, e.g., in Germany “Institut für Bautechnik (DIBt)”.2 But there is a large grey market for wood preservatives other than the authorized substances. As a result, a large variety of solvents can occur in wood preservatives. Different systems of classification are used worldwide. In Germany, authorized wood preservative substances are published in an index of wood preservatives (“Holzschutzmittelverzeichnis”)2 which is elaborated by the DIBt and the central German environmental authority (“Umweltbundesamt”). A systematic survey of wood preservatives is shown in Table 14.14.1, including wood preservatives containing solvents. Solvents are normally found in wet systems of wood preservatives. Commonly used solvents are substances which are applied in connection with normally used binders (aldehyde resins, acrylates and polyurethanes). Water or the appropriate solvents are added to binders. Table 14.14.1. Systems of wood preservatives Purpose and base

Terms “CF-salts” “CFA-salts”

Water-soluble agents as preventive treatment against fungi and insects

“SF-salts” “HF-salts” “B-salts” “Single CK-salts” “CKA-salts” “CKB-salts” “CKF-salts” “CFB-salts” collective group

Oily agents as preventive treatment against fungi and insects

tar oil preparations preparations containing solvents pigment-free preparations containing binders and solvents preparations with stained pigments containing solvents special preparations only used in stationary installations preparations containing coal tar oil

Preparations used for special applications

Active components chromium and fluorine compounds alkali fluorides, alkali arsenate, and bichromate (no longer permitted) silicofluorides hydrogen fluorides inorganic boron compounds copper salts, bichromate copper salts, bichromate with arsenic compounds copper salts, bichromate with boron compounds copper salts, bichromate with fluorine compounds chromium, boron and fluorine compounds other compounds, e.g., bis(N-cyclohexyldiazeniumdioxyl)-copper distillates of bituminous coal tar (carbolineum) organic fungicides and insecticides organic fungicides and insecticides organic fungicides and insecticides organic fungicides and insecticides organic agents, special distillates containing coal tar oil, solvents and pigments

pastes wood preservatives used in particle board in manufacturing plants agents used as preventive treatment against insects contain organic insecticides

14.15 Medical applications

955

Solventborne wood preservatives contain mainly nonpolar, organic solvents apart from other substances such as fungicides and insecticides.1 These solvents are classified as VOCs. 14.14.2.2 Technical and environmental aspects Solvent-based wood preservatives show several advantages, especially in their application and technical effectiveness.1 They can be applied repeatedly and do not alter the structure of the wood products. Application is faster and the characteristics of the final product are improved, e.g., visual appearance of surface. Nevertheless, there are some disadvantages, especially environmental ones. Most solvents are released quickly (VOCs) and can cause severe environmental effects. This is especially true if toxic solvents are employed. Emissions of solvents from wood products are described under various conditions, e.g. indoor air emissions from furniture or emissions in test chambers.3,4 Solvents can be emitted as primary or reactive products of the wood product or the coating system; solvents can also be investigated as secondary emission products.3,5 The emission characteristics depend on solvent properties and surrounding conditions, e.g., air velocity and air exchange rate.6 In the indoor air, solvents from wood products follow various pathways. Examples of interactions are possible reactions of solvents (e.g., styrene) with air components (e.g., hydroxy radicals),3 transport into and through indoor materials7 or sorption processes.5 The emitted solvents can be reduced by ventilation processes or they may be absorbed by organisms. For humans, absorption of the wood preservatives or ingredients (e.g., solvents) can cause various toxic effects. It is often difficult to pinpoint the causative agents (see Chapter 20). REFERENCES 1 2 3 4 5 6 7

Ullmann`s Encyclopedia of Industrial Chemistry, 1998. DIBt (Deutsches Institut für Bautechnik). Holzschutzmittelverzeichnis. Index of wood preservatives (1999). T. Salthammer, A. Schwarz, F. Fuhrmann, Atmospher. Environ., 33, 75 (1999). T. Salthammer, Atmospher. Environ., 7, 189 (1997). M. Wensing, H.J. Moriske, T. Salthammer, Gefahrstoffe Reinhaltung der Luft, 58, 463 (1998). E. Uhde, A. Borgschulte, T. Salthammer, Atmospher. Environ., 32, 773 (1998). R. Meininghaus, T. Salthammer, H. Knoppel, Atmospher. Environ., 33, 2395 (1999).

14.15 MEDICAL APPLICATIONS George Wypych ChemTec Laboratories, Inc., Toronto, Canada

Industries manufacturing medical devices use a wide variety of technological processes which most likely take advantage of most of the available solvents. The range of solvent use is so wide that a complete description of each solvent and its application is not possible in this book. It is questionable if such analysis is possible given that many processes are guarded by trade secrets where there is no patent disclosure. Some examples are given,

956

George Wypych

more to show that, although solvents do contribute to pollution, they also help to produce materials which are needed for health and well being. Polyurethanes are materials which have the required properties and biocompatilibity which makes them good candidate for use in medical devices. Common applications include pacemaker leads, peripheral and central catheters, feeding tubes, balloons, condoms, surgical gloves, instrument and appliance covers, wound dressings, and many other.1-4 Several methods are used to process polyurethanes. These include injection molding, extrusion, and solution processing. In solution processing film casting and dip molding are the most frequent techniques. Dimethylacetamine, tetrahydrofuran, dichloromethane, methyl ethyl ketone, N,N-dimethylformamide, N-methylpyrrolidone, cyclopentanone, cyclohexanone, dioxane, and chloroform are the most commonly used solvents. Most of these are hazardous but used because they contribute to highly transparent product which is very desirable in medical devices. Transparent materials can only be made from transparent solutions.1 These solvents can dissolve polymers well and form clear solutions. Ease of solvent removal from the material is very important in formulation design. Obviously, no traces of solvents should remain in the medical devices since even trace amounts may interfere with the treatment and the patient’s health. An inappropriate solvent selection may cause the formation of crust as the solvent escapes. This leads to material discontinuity (e.g., pinholes) which renders the product inferior. This brings a discussion of solvent evaporation, the rheological properties of formulation, and formation of multilayer materials. Good solvents can be used in lower concentration but they result in viscous solutions which, in dip coating, form thick films which have the potential of blistering on evaporation. If the solution is diluted, film continuity suffers which increases the number of pinholes. Rapid evaporation causes a formation of a crust of gelled solidified polymer which makes solvent removal more difficult and damages the integrity of the layer. Also, material does not have time to adjust and leveling suffers. On the other hand slow evaporation may cause dissolution of the layer below the coating in a multilayered products and bubbling between the layers. The selection of solvents for dip coating is usually a complex process ultimately requiring multicomponent solvent mixtures which include a good solvent, a poor solvent, and a solvent of lower boiling point (sometime called “blush resistor”) to balance viscosity and rate of evaporation.2 In wound dressings, the solvents selected affect the material microstructure which controls the evaporation of exuded body fluids but prevents bacteria and pathogens from entering the wound.3 In infection-resistant medical devices, the antimicrobial agent must be uniformly distributed over all areas of the medical device which may come into contact with a patient. Otherwise there is a risk of infection.4 Not all solvents dissolve antimicrobial agents and swell surface of medical device. Cleaning of penetrable septa, tubing systems, and infusion and dialysis systems is another application in which solvents are used. The solvents which are suitable for elastomer cleaning are dichloromethane, perchloroethylene, halogenated hydrocarbons, and freons.5 This cleaning method extracts undesirable organic materials from medical devices which might otherwise be extracted by body fluids. Heat treatment of catheters followed by washing with a polar solvent increase its surface lubricity. Catheter with poor surface lubricity often causes frictional pain upon its insertion into the body cavity and damages the mucosal tissue resulting in cross infection.6 Film dressings contain two types of solvents: solvents to

14.16 Metal casting

957

dissolve the polymer and propellant solvents. These must be selected to achieve technological goals related to solubility and compatibility.7 These examples show that the many technological considerations place constraints on in solvent selection. Solvent replacement in complex products and technological processes is a long-term, expensive proposition which usually results in a need for complete reformulation of the material with failure to achieve the objective a very possible outcome.


A J Walder, Plast. Eng., 54, No.4, 29-31 (1998). M T Shah, US Patent 5,571,567, Polygenex International, Inc., 1996. J Delgado, R J Goetz, S F Silver, D H Lucast, US Patent 5,614,310, 3M, 1997. S Modak, L Sampath, US Patent 5,567,495, Columbia University, 1996. S H Smith, J M Brugger, H W Frey, US Patent 5,639,810, COBE Laboratories, Inc., 1997. L Mao, Y Hu, D Piao, US Patent 5,688,459, China Rehabilitation Research Center, 1997. A J Tipton, S M Fujita, R L Dunn, US Patent 5,792,469, Atrix Laboratories, Inc., 1998.

14.16 METAL CASTING George Wypych ChemTec Laboratories, Inc., Toronto, Canada

The metal casting industry has 3,100 facilities in the USA and employs 250,000 people. Most plants are small and technological processes are very diverse. The processes do share common phases, including pattern making, mold and core preparation, furnace charge and metal melting, mold charging, cooling and finishing. Most steps use solvents. In the mold-making process, a many chemical binding systems are used, some of which contain methanol, benzene, toluene, and cresol. The metal is most often recycled and it typically requires cleaning before it is charged to the furnace. This is accomplished either by precombustion or solvent cleaning. In die casting operations, solvent-based or water-based lubricants are used. Die casters also use die fluxes which contain solvents. Some solvent replacement additives in water-based lubricants contain hazardous solvents. Finishing operations involve casting cleaning to remove scale, rust, oxides, oil, grease, and dirt. Solvents are typically chlorinated solvents, naphtha, toluene, and methanol. Cleaning can also be done by emulsifiers, abrasives, alkaline agents, and acid pickling. The cleaning operation is usually followed by painting which frequently involves solvent-based paints and thinners. Tables 14.16.1 and 14.16.2 contain information on the reported solvent releases and transfers from metal casting industry. The data show that solvent use is not excessive relative to other industries. The industry plans to further improve its environmental record by developing environmentally improved materials which meet regulations. The solvent cleaning and die lubrication are processes under study.

REFERENCES 1 2

EPA Office of Compliance Sector Notebook Project. Profile of the Metal Casting Industry. US Environmental Protection Agency, 1998. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.

958

George Wypych

Table 14.16.1. Reported solvent releases from the metal casting industry in 1995 [Data from Ref. 1] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

110,000

methyl ethyl ketone

22,000

n-butyl alcohol

15,000


22,000

cresol

20,000


41,000

dichloromethane

50,000

tetrachloroethylene

13,000

ethylbenzene

10,500


111,000

ethylene glycol

64,000

trichloroethylene

75,000

hexachloroethane

16,000

toluene

233,000

methanol

5,860,000

xylene

388,000

Table 14.16.2. Reported solvent transfers from the metal casting industry in 1995 [Data from Ref. 1] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

115


22,000

ethylbenzene

340


500

ethylene glycol

50,000

trichloroethylene

1,000

methanol

10,000

toluene

4,000

methyl ethyl ketone

8,000

xylene

82,000

14.17 MOTOR VEHICLE ASSEMBLY George Wypych ChemTec Laboratories, Inc., Toronto, Canada

Automotive industry in US alone employs 6.7 million person and it is a large contributor to the gross national product. It uses large quantities of solvents and it is perceived to contribute to pollution by solvents and other materials.1-6 Solvents are used in a variety of cleaning, preparation, and painting operations. Automotive finishing process may be divided into four main categories: anti-corrosion operations (cleaning, phosphate treatment, and chromic acid treatment), priming operations (electrodeposition of primer, anti-chip coating application, and primer application), joint sealant application, and other finishing operations (color coat, clear coat). These main operations employ many materials which contain solvents.

14.17 Motor vehicle assembly

959

The cleaning process involves acid/alkaline and solvent cleaning. Typical solvents involved are acetone, xylene, toluene, and 1,1,1-trichloroethylene. The primer bath is water-based but usually some organic solvents are present (5-10%). These solvents are the same as those listed above. After the application of primer, the car body is baked and then undergoes waterproofing with an application of polyvinylchloride sealant which contains a small amount of solvents. Following waterproofing, the automotive body proceeds to the anti-chip booth, where urethane or epoxy solvent-coating systems are applied. This process is followed by application of primer-surfacer coating which is either a polyester or an epoxy ester in a solvent system. The primer-surfacer coating is applied by spraying and provides a durable finish which can be sanded. After the sanding step, the primary color coating is applied also by spraying. These primary color formulations contain about twice as much solvent as the primer-surfacer coating. Solvents are flashed-off (no heating) and a clear coat is applied. Then the entire car body is baked for about 30 min. Solvents used include butanol, isobutanol, methanol, heptane, mineral spirits, butyl acetate ethyl acetate, hexyl acetate, methyl ethyl ketone, acetone, methyl amyl ketone, toluene, and xylene. Several finishing operations also employ solvents. After baking, a sound-deadener is applied to certain areas of the underbody. It is a solvent based material with a tar-like consistency. A trim is applied with adhesives which contain solvents (see section on adhesives and sealants). After the installation of trim and after the engine is installed, car undergoes an inspection. Some repainting is required in about 2% of the production. If damage is minor then repainting is done by a hand operated spray gun. If the damage is substantial a new body is installed. Equipment cleaning solvents are also used. Spraying equipment is cleaned with a “purge solvent” which may consist of a mixture of dimethylbenzene, 4-methyl-2-pentanone, butyl acetate, naphtha, ethyl benzene, 2-butanone, toluene, and 1-butanol. Tables 14.17.1 and 14.17.2 contain information on the reported solvent releases and transfers from the motor vehicle assembly industry. The data show that solvent use is very large compared with all industries covered so far in our discussion except for the steel and iron industry. The motor vehicle assembly industry is the sixth largest producer of VOC and also the sixth largest industry in reported emissions and transfers. The data in Tables 14.17.1 and 14.17.2 are data from 1995 the most recent available. The automotive industry and associated paint companies conduct extensive work on replacement of VOC containing paint systems. These efforts are mainly directed to water-based systems and powder coatings. Until recently, water-based systems were preferred but now attention is shifting to powder coatings which eliminate VOC. There is no status quo. Changes are dynamic and kept protected by trade secrets which makes it difficult to comment on specific progress. Solvent use by the European industry5 is that the production of one car requires an average of 10 kg of solvents. Solvents use is not the only problem the industry is facing. 16% of the total energy used in car production is required by painting and finishing operations. Both energy conservation and reduction is solvent consumption must be pursued to meet environmental objectives. Not only can these issues be addressed through material reformulation but the design of equipment used in applying and drying the coating can also reduce emission and save energy. A new trend is apparent as plastics are introduced to automotive production. Plastic parts must also be painted. Paint systems are difficult to select. Chlorinated polyolefins provide good adhesion of paints and reduce VOC but are also under scrutiny because of pres-

960

George Wypych

Table 14.17.1. Reported solvent releases from the motor vehicle assembly industry in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

13,000

methyl ethyl ketone

2,320,000

n-butyl alcohol

2,260,000


3,060,000

sec-butyl alcohol

86,000


193,000

tert-butyl alcohol

4,200

methyl tert-butyl ether

32,000

cyclohexane

35,000

tetrachloroethylene

140,000

dichloromethane

380,000


730,000

ethylbenzene

1,370,000

trichloroethylene

1,300,000

ethylene glycol

180,000


1,120,000

isopropyl alcohol

9,000

toluene

2,610,000

hexane

95,000

xylene

10,800,000

methanol

1,550,000

m-xylene

25,000

Table 14.17.2. Reported solvent transfers from the motor vehicle assembly industry in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

3,400

methyl ethyl ketone

2,100,000

n-butyl alcohol

1,030,000


4,700,000

sec-butyl alcohol

9,000


330,000

tert-butyl alcohol

1,000


2,300

cyclohexane

670

tetrachloroethylene

49,000

dichloromethane

450,000


140,000

ethylbenzene

1,740,000

trichloroethylene

480,000

ethylene glycol

605,000


330,000

isopropyl alcohol

2,000

toluene

2,020,000

hexane

25,000

xylene

9,200,000

methanol

760,000

m-xylene

2,100

ence of chlorine. Powder coatings are available7 but they require a high energy input. These problems are apparent but the solution to them will take several years to implement due, in large part, to the long term testing needed to confirm coating performance (up to 5 years in Florida).

14.18 Organic chemical industry

961

Table 14.18.1. Reported solvent releases from the organic chemical industry in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

allyl alcohol

31,000

ethylbenzene

370,000

benzene

690,000

ethylene glycol

6,050,000

n-butyl alcohol

850,000

hexane

600,000

sec-butyl alcohol

63,000

isopropyl alcohol

150

tert-butyl alcohol

430,000

methanol

8,750,000

carbon disulfide

85,000

methyl ethyl ketone

260,000


10,000


520,000

chlorobenzene

27,000


350,000

chloroform

7,000


64,000

cresol

280,000

pyridine

120,000

m-cresol

320,000

tetrachloroethylene

20,000

o-cresol

270,000

toluene

1,040,000

p-cresol

162,000

1,2,4-trichlorobenzene

41,000

cyclohexane

450,000


130,000

cyclohexanol

1,100,000

trichloroethylene

18,000

dichloroethane

120,000

xylene

350,000

1,2-dichloroethylene

70

m-xylene

59,000

dichloromethane

310,000

o-xylene

34,000


25,000

p-xylene

660,000

1,4-dioxane

12,000


EPA Office of Compliance Sector Notebook Project. Profile of the Motor Vehicle Assembly Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998. G. Wypych, Ed., Weathering of Plastics. Testing to Mirror Real Life Performance, Plastics Design Library, Society of Plastics Engineers, New York, 1999. M Harsch, M Finkbeiner, D Piwowarczyk, K Saur, P Eyerer, Automotive Eng., 107, No.2, 211-4 (1999). C A Kondos, C F Kahle, Automotive Eng., 107, No.1, 99-101 (1999). D C Shepard, J. Coat. Technol., 68, No.857, 99-102 (1996). T Hosomi, T Umemura, T Takata, Y Mori, US Patent 5,717,055, Mitsubishi Gas Chemical Company, Ltd., 1998.

962

George Wypych

Table 14.18.2. Reported solvent transfers from the organic chemical industry in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

allyl alcohol

210,000

ethylbenzene

980,000

benzene

420,000

ethylene glycol

6,800,000

n-butyl alcohol

1,500,000

hexane

770,000

sec-butyl alcohol

1,700,000

isopropyl alcohol

85,000

tert-butyl alcohol

12,500,000

methanol

23,000,000

carbon disulfide

96,000

methyl ethyl ketone

800,000


12,000


390,000

chlorobenzene

130,000


110,000

chloroform

92,000


210,000

cresol

430,000

pyridine

33,000

m-cresol

720,000

tetrachloroethylene

138,000

o-cresol

57,000

toluene

4,400,000

p-cresol

870,000

1,2,4-trichlorobenzene

8,000

cyclohexane

900,000


290,000

cyclohexanol

3,700

trichloroethylene

42,000

dichloroethane

230,000

xylene

4,000,000

1,2-dichloroethylene

1,000

m-xylene

51,000

dichloromethane

870,000

o-xylene

460,000


370,000

p-xylene

1,700

14.18 ORGANIC CHEMICAL INDUSTRY George Wypych ChemTec Laboratories, Inc., Toronto, Canada

The chemical industry operates about 1000 plants in the USA with 53 companies producing 50% of the total output of $65 billion in sales and employing 125,000 people. There are point source solvent emissions (e.g., laboratory hoods, distillation units, reactors, storage tanks, vents, etc.), fugitive emissions (e.g., pump valves, flanges, sample collectors, seals, relief devices, tanks), and secondary emissions (waste water treatment units, cooling towers, spills). Organic liquid wastes containing solvent are generated from processes such as equipment washing, surplus chemicals, product purification, product reaction, housekeeping, etc.

14.19.1 Architectural surface coatings

963

Tables 14.18.1 and 14.18.2 give the reported solvent releases and transfers from the organic chemical industry. Large quantities of solvents are involved. The organic chemical industry produced the second largest quantity of VOC and the second largest releases and transfers. The industry is actively working to reduce solvent use because of the high costs (waste treatment, fines, liabilities, etc). There are many efforts under way to reduce environmental emissions and improve safe practices. The initiatives include process modifications such as a reduction in non-reactive materials (e.g., solvents) to improve process efficiency, a reduction in the concentration of chemicals in aqueous solution, and improved R&D and process engineering. Equipment modifications are planned to reduce leaks, prevent equipment breakdown, and improve the efficiency of emission control devices.

REFERENCES 1 2

EPA Office of Compliance Sector Notebook Project. Profile of the Organic Chemical Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.

14.19 PAINTS AND COATINGS 14.19.1 ARCHITECTURAL SURFACE COATINGS AND SOLVENTS

Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg Institut fòr Allgemeine Hygiene und Umwelthygiene Universit¬t Tòbingen, Tòbingen, Germany


14.19.1.1 General aspects Coating materials and coating techniques can be distinguished and systematized in various ways. The fundamental principles of common coating systems are:1 • Physical drying. A solid surface film is formed after the evaporation of water or organic solvents. • Physico-chemical drying/curing. Polycondensation or polyaddition are combined with evaporation of organic solvents. • Chemical curing. Solvents, e.g., styrene or acrylic monomers, react with the curing system. The actual effects depend on the surrounding conditions and the ingredients of the coating system, e.g., solvents. Solvents contribute many essential properties to coating systems. Solvents can improve technical factors such as application or surface properties. Solvents also bring negative qualities to coating materials, especially with respect to environmental conditions (e.g., toxic effects of emitted organic solvents). 14.19.1.2 Technical aspects and properties of coating materials Application techniques for coatings can be considered in various ways. The stability and durability of coating is essential. Coatings that have normal wear and tear requirements are

964


mainly based on oils and aldehyde resins. Higher durability or stability can be achieved by the use of one of the following one- or two-component systems. One-component: • Bituminous materials • Chlorinated rubber • Polyvinyl chloride • Polyacrylic resin • Polyethene • Saturated polyester • Polyamide Two-component: • Epoxy resin • Polyurethane • Mixtures of reactive resins and tar A survey of the performance of different coating materials together with an assessment of various environmental factors is given in Table 14.19.1.1. Table 14.19.1.1. Environmental performance of some coating materials

Abbreviation

Weathering response

Acid atmosphere

Humidity

Under water

acid

alkali

Solvent

Temperature 275°C

Not a VOC

Evaporation rate

< 0.01 (Butyl acetate = 1.0)

Slow, allowing excellent coalescence

Freezing point

< -55°C

No freeze-thaw issue

Water solubility

600 ppm

Hydrolytic stability

Very Good

No hydrolysis in normal use

Color

5 Hazen units

Imparts no color to coatings

Coalescing efficiency

Excellent for most polymer systems

Biodegradability

80% in 28 days

Odor

None discernible

Toxicology

Oral LD50 (Rat) >16,000 mg/kg

Biodegradable

Essentially non toxic

14.19.2.7 RECENT ADVANCES IN DIESTER COALESCING SOLVENTS We have also prepared the di-isopropyl esters of the higher adipic content stream. This has a vapor pressure similar to that of Coasol, but is slightly more water soluble. Finally, we have manufactured di-isopropyl adipate, which has the highest boiling point, the lowest vapor pressure and the lowest water solubility of all of this range of products. These preparations were undertaken to add to repertoire of products to suit the diverse requirements of the formulators of aqueous based systems. In virtually all cases, the dibasic esters gave a significant improvement in efficiency in reducing the MFFT for a given quantity of additive. The attempts are made to offer a tailor made solution to each individual polymer system employed in the development of aqueous based systems. The dibasic esters of the AGS acids group offer the opportunity for fine tuning, with the added advantage of low odor, low toxicity and “excellent” VOC status.

14.20 Petroleum refining industry

975

14.19.2.8 Appendix - Classification of coalescing solvents Coalescent Type Type A

Type of Species

Examples

Comments

Hydrocarbons

White Spirit

Type AB

Diesters

DBE Dimethyl esters DBE Diisobutyl esters Di-isobutyl adipate Di-isopropyl adipate Dibutyl phthalate

Estasol, Du Pont DBE’s Coasol, Lusolvan Chemoxy new products

Type AB

Ester alcohols

Diol Monoesters

Texanol

Type ABC

Glycol esters & Glycol ester ethers

PGDA Butyldiglycol acetate

Type ABC

Ether alcohols & diethers

PnBS 2-Butoxyethanol MPG Diethers

Type C

Glycols

DEG DPG TEG

Dow Products BASF and others Proglides and glymes

14.20 PETROLEUM REFINING INDUSTRY George Wypych ChemTec Laboratories, Inc., Toronto, Canada

The US petroleum refining industry generates sales of over $140 billion with only about 200 plants. It employs 75,000 people. About 90% of the products used in US are fuels of which 43% is gasoline. Figure 14.20.1 illustrates how the products breakdown. The process is described in detail in Chapter 3. Emissions of hydrocarbons to the atmosphere occur at almost every stage of the production process. Solvents are produced in various processes and they are also used to extract aromatics from lube oil feedstock, deasphalting of lubricating base stocks, sulfur recovery from gas stream, production of solvent additives for motor fuels such as methyl tert-butyl ether and tert-amyl methyl ether, and various cleaning operations. Emissions to atmosphere include fugitive emissions of the volatile components of crude oil and its fractions, emissions from incomplete combustion of fuel in heating system, and various refinery processes. Fugitive emissions arise from thousands of valves, pumps, tanks, pressure relief valves, flanges, etc. Individual leaks may be small but their combined quantity results in the petrochemical industry contributing the largest quantity of emissions and transfers. Tables 14.20.1 and 14.20.2 give solvent releases and transfers data for the petroleum refining industry. Transfers are small fraction of releases which means that most wastes are processed on-site. In addition to emissions to atmosphere, some plants have caused contamination of ground water by releasing cooling and process water.

976

George Wypych

Toluene, xylenes, and benzene constitute the majority of solvent emissions since they are native components of crude oil. Methyl ethyl ketone is also emitted in large quantities because of its use in lube oil dewaxing. Pollution prevention will become increasingly important to the petroleum industry as federal, state and municipal regulations become more stringent and waste disposal cost rises. The industry estimates that to comply with 1990 Clean Air Act Amendments it will require investment of $35-40 billion. Actions required to decrease pollution include process equipment modification, waste segregation and separation, recycling, and better training and supervision.

Figure 14.20.1. Diagram of production outputs from refineries. [Reproduced from EPA Office of Compliance Sector Notebook Project. Profile of the Petroleum Refining Industry. US Environmental Protection Agency, 1995.

Table 14.20.1. Reported solvent releases from the petroleum refining industry in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

1,750,000


110,000

n-butyl alcohol

23,000


280,000

tert-butyl alcohol

28,000

methyl tert-butyl ketone

1,380,000


17,000

tetrachloroethylene

21,000

cresol

75,000


50,000

cyclohexane

960,000

trichloroethylene

730

dichloromethane

8,000


420,000

ethylbenzene

600,000

toluene

4,360,000

ethylene glycol

46,000

xylene

2,330,000

hexane

3,000,000

m-xylene

170,000

methanol

540,000

o-xylene

150,000

methyl ethyl ketone

2,100,000

p-xylene

1,000,000

14.21.1 Use of solvents in manufacture of drugs

977

Table 14.20.2. Reported solvent transfers from the petroleum refining industry in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

160,000


4,000

tert-butyl alcohol

900

methyl tert-butyl ketone

34,000


1,000

tetrachloroethylene

900

cresol

130,000


6,500

cyclohexane

10,000


31,000

ethylbenzene

61,000

toluene

270,000

ethylene glycol

58,000

xylene

340,000

hexane

13,000

m-xylene

11,000

methanol

180,000

o-xylene

30,000

methyl ethyl ketone

30,000

p-xylene

7,000


3,500

REFERENCES 1 2

EPA Office of Compliance Sector Notebook Project. Profile of the Petroleum Refining Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.

14.21 PHARMACEUTICAL INDUSTRY 14.21.1 USE OF SOLVENTS IN THE MANUFACTURE OF DRUG SUBSTANCES (DS) AND DRUG PRODUCTS (DP)

Michel Bauer International Analytical Department, Sanofi-Synthélabo, Toulouse, France

Christine Barthélémy Laboratoire de Pharmacie Galénique et Biopharmacie, Faculté des Sciences Pharmaceutiques et Biologiques, Université de Lille 2, Lille, France

14.21.1.1 Introduction Today the manufacturing of a pharmaceutical drug is almost totally the responsibility of: • the chemical industry for the preparation of the drug substance (active principle) and the excipients used for preparing the DP (finished product) • the pharmaceutical industry for the preparation of the DP itself. For the reader who is not familiar with the pharmaceutical industry, a reminder of the key points is given below.

978

Michel Bauer, Christine Barthélémy

One or several active compounds (DS) are prepared by organic synthesis, extracted from vegetable, animals, micro-organisms or obtained by biotechnology. The DS is generally associated in the product with several excipients of chemical, mineral or biological in nature either as monomers or as polymers. The goal is to formulate a processable DP stable over the time and allowing the active substance to be released in vitro and in vivo. Obviously the formulation is designed in relation to the route of administration: • oral route solid dosage forms (e.g., tablets, capsules, etc.) drinkable solutions etc. • ORL route (nasal solutions, spray) • local route (suppositories, transdermal systems, eye-drop formulation, spray) • intravenous and intramuscular route (injectable solution, lyophilizate, etc.) At practically every step of the manufacture of the drug substance and the excipients, solvents including water are utilized. This is equally true for the preparation of the pharmaceutical formulations. Ideally we would like to have available a universal stable solvent, ultrapure, non-toxic and which does not affect the solutes. This is an old dream of the alchemists who searched for a long time for the “Alkahest” or “Menstruum universal” as it was named by Paracelsus.1 The fact that a solvent is not a totally inert species allows it to play an important role in chemical equilibria, rates of chemical reactions, appearance of new crystalline forms, etc. and consequently contributes to the great wealth of compounds which the chemists are able to produce. But of course, there are drawbacks in using solvents. Because there are not totally inert they may favor the formation of undesirable impurities in the intermediates of synthesis and in the DS. Regarding the manufacture of the DP, the solvents, including water, may induce either polymorphic transformations or formation of solvates (hydrates) which, after drying, could lead to a desolvated solvate with quite different physical properties impacting potentially either positively or negatively on the DP performance.2 Another crucial aspect which deserves to be discussed, is the notion of purity. Impurities present in solvents could have an impact on the stability of drugs, for example, or on the crystallization process. Last, but not least, the toxicological aspects should be taken into account. Numerous solvents show different kinds of toxicity and this should be a matter of concern in relation to the health of workers exposed to them.3 But ultimately residual solvents still present in the DS and DP have to be assessed and systematically limited. We are now going to consider several aspects of the use of solvents in the manufacture of drug substances (DS) and drug products (DP) including their quality (purity) and influence on the quality, stability and physico-chemical characteristics of pharmaceutical products. The issue of residual solvents in pharmaceutical products will be considered in Chapter 16.2 and will focus amongst other things on the corresponding ICH Guideline.4


979

14.21.1.2 Where are solvents used in the manufacture of pharmaceutical drugs? 14.21.1.2.1 Intermediates of synthesis, DS and excipients 14.21.1.2.1.1 General points Raw materials are now produced by the chemical industry and involve the use of solvents at different steps in their production. These materials are usually produced by: • chemical synthesis • an extraction process, a fermentation process • or a biotechnology process The goal of this chapter is, of course, not to deal with the criteria for selection in relationship to their use in particular chemical reactions or extraction processes but rather to stress that impurities present in solvents could have an impact on the purity of the substances obtained, on their stability and potentially on their safety. These three concepts are of paramount importance in the pharmaceutical field. A list of solvents which are commonly used in the chemical industry5 is presented in Table 14.21.1.1. Table 14.21.1.1. Solvents commonly used in the chemical industry

Alcohols

Ketones

Ethanol

Acetone

Butanol

Methyl ethyl ketone

2-Ethylhexanol


Isobutanol

Methyl isopropyl ketone

Isopropanol

Mesityl oxide

Ethylene bromide Chloroform Ethylene chloride Dichloromethane Tetrachloroethylene Carbon tetrachloride

Methanol Propanol

Ethers

Propyleneglycol

1.4-Dioxane Butyl ether

Amide

Ethyl ether

Dimethylformamide

Diisopropyl ether Tetrahydrofuran

Amine

Halogenated solvents

Tert-butyl methyl ether (MTBE)

Trichloroethylene

Sulphur-containing Dimethylsulfoxide

Aromatics hydrocarbons Toluene Xylene

Pyridine Nitriles Aliphatic hydrocarbons

Acetonitrile

Ethyl acetate

Cyclohexane Hexane

Esters

Water

It is generally relatively easy to know for pharmaceutical industry the nature of solvents to be looked for in a DS because it produces itself the active component or because it

980


can have by contract an access to the DS Mater File or because there is a compendial monograph giving occasionally some indications (e.g., search for benzene in carbomers). As a consequence of the ICH Guideline Q3C4 dealing with the residual solvents in pharmaceutical products (see Chapter 16.2), the use of solvents of class I (solvents to be avoided) like benzene is no more possible. It is known that carbopol resins (carbomer), used to modify the rheology of polar systems and as a binder in sustained release tablets, were up to now polymerized in benzenic medium. In the current quality of poloxamers it was possible to retrieve up to 1000 ppm of benzene. The ICH limit being 2 ppm, it was impossible to achieve this goal. The manufacturers have consequently developed new polymerization media6 containing either ethyl acetate alone or a mixture of ethyl acetate and cyclohexane, the first one belonging to class III (no safety concern), the second one belonging to class II (ICH limit 3880 ppm). 14.21.1.2.1.2 Criteria of purity This is a difficult matter. Purity in chemistry is an ideal concept referring to a situation where a product consists of one type of molecules only. This is a theoretical situation which can only be EXPERIMENTALLY approached more or less closely.7 The purity of a product is a relative notion and is dependant on the analytical methods used and their performances. More practically the quality finally chosen for a solvent will depend on the specific use for which this solvent is intended to be utilized.8 A solvent is considered sufficiently pure if it does not contain impurities able in nature and in quantity to interfere on the admissible quality of the product in the manufacture of which it participates.7 14.21.1.2.1.3 Solvents as reaction medium In this case the solvents should have a range between the melting point and the boiling point as extended as possible and a good thermal and chemical stability. The purity should be of good degree but could depend on the step considered of the global synthesis. As an example let us consider the case of the dimethylformamide (DMF).5 If it is used in reactions evolving in anhydrous media, it will be mandatory to control the level of water at the ppm level. The specification regarding the water content will be of course loosened if the DMF in the chemical step considered is used in conjunction with water as reaction media. We will see further that solvents contain actually a lot of chemical impurities which could be reactive vis a vis the main molecule undergoing the chemical reaction and leading to additional impurities other than those coming from the mechanism of reaction itself. 14.21.1.2.1.4 Solvents for crystallization They should be carefully chosen in such a way that they show a high solubility at high temperature and a low solubility at low temperature of the substance to be crystallized or recrystallized. Of course the solvent should be absolutely inert and of the highest achievable purity for at least two reasons: • The first one being identical to the one mentioned for solvents as reaction media: possibility to produce other impurities. • The second one being linked to the crystallization process itself. It is well known that the presence of impurities whatever the origin could have serious effects on the nucleation and growth process. We will tell a little bit more about that further in the text.


981

14.21.1.2.1.5 Solvents used for extraction and preparative chromatography As in the precedent cases they have to be absolutely inert (as far as it is possible) and with a high degree of purity for the reasons already evoked. In case of preparative chromatography a special care will be taken concerning the chemical inertia to adsorbate9 of the solvents constituting the mobile phase and the fact that impurities or additives contained in the solvents in a way not under control could impair significantly the reproducibility of the retention times. 14.21.1.2.1.6 Nature and origin of impurities contained in solvents10 It should be reminded here that a solvent used at the industrial level is rarely pure (we mean here no impurity analytically detectable). Industrial solvents may contain: • impurities coming from their origin or their manufacturing process • impurities originating from the container during transportation • stabilizers • denaturing agents • impurities resulting from a transformation of the solvent during the chemical reaction These impurities or side products should be look for as far as it is possible when assessing the purity of the solvent. In fact they could be less volatile than the main solvent and could finally concentrate in the pharmaceutical product. We will now review shortly the nature of all these kinds of impurities of the most often used solvents. 14.21.1.2.1.6.1 Impurities coming from the origin or the manufacturing process of the solvent1,10 Table 14.21.1.2. Solvent impurities Class of solvents

Possible impurities (according to the manufacturing process) Hydrocarbons

Toluene

Methylthiophene, benzene, paraffinic hydrocarbons

Xylene

Mixture of ortho, meta and para isomers, paraffinic hydrocarbons, ethyl benzene, sulfur compounds

Cyclohexane

Benzene, paraffinic hydrocarbons, carbonyl compounds Halogenated compounds

Dichloromethane

Chloroform, carbon tetrachloride, chloromethane

Chloroform

Chlorine, carbonyl chloride (phosgene), dichloromethane, carbon tetrachloride, hydrogen chloride


Chlorides, chlorine, carbon disulfide Alcohols

Methanol

Water, acetone, formaldehyde, ethanol, dimethylether, carbon dioxide, ammonia

methyl

formate,

982


Ethanol

Aldehydes, ketones, esters, water, ethyl ether, benzene (if anhydrous ethanol)

2-propanol

Water, peroxides

N.B.: Some alcohols obtained by fermentation could contain pesticides. It is necessary to obtain from the purchaser some guaranty in requiring limit contents (expressed in Parathion e.g.). Aliphatic ethers/cyclic ethers Ethylether/isopropyl ether/monoalkylated ethers/ethylene glycol/diethylene glycol/etc.

Alcohols (from which they are prepared), water, corresponding aldehydes, peroxides

Tetrahydrofuran

Water, peroxides

Dioxane

Acetaldehyde, water, acetic acid, glycol acetal paraldehyde, crotonaldehyde/peroxides Ketones

Acetone

Methanol, acetic acid, water Esters

Methyl acetate

Acetic acid, water, methanol

Ethyl acetate

Acetic acid, ethanol, water Amides

Formamide

Formic acid, ammonium formate, water


N-Methylformamide, formic acid, water Nitriles

Acetonitrile

Acetamide, ammonium acetate, ammoniac, water, toluene Nitro compounds

Nitrobenzene

Nitrotoluene, dinitrothiophene, dinitrobenzene, aniline

14.21.1.2.1.6.2 Impurities originating from the container during transportation It relates to contamination coming from tankers or drums not correctly cleaned. These concerns of course solvents of low quality conveyed in industrial quantity. In case of utilization of such solvents, the user has to bear in mind that some incidents or uncommon behavior may find an explanation based on this considerations. 14.21.1.2.1.6.3 Stabilizers It is of course very difficult to know every stabilizer used. There is here an important problem of confidentiality. We quote thereafter some of them which are well known.


983

Table 14.21.1.3 Stabilizers used in selected solvents Solvents

Stabilizers

Dichloromethane

Ethanol, 2-methyl-but-2-ene

Chloroform

Ethanol (1% V/V) for avoiding the phosgene formation, 2-methyl-but-2-ene

Diethylether

2,6-di-tert-butyl-4 methylphenol (BHT)

Tetrahydrofuran

BHT, p-cresol, hydroquinone, calcium hydride

14.21.1.2.1.6.4 Denaturing agents This process is relevant primarily to ethanol. Common denaturing agents are: methanol, isopropanol, ethyl acetate, toluene. 14.21.1.2.1.6.5 Transformation of the solvent during the chemical reaction Solvents are rarely chemically inert. During the reactions where solvents are involved, they can undergo chemical transformation generating impurities which can be found, for example in the DS. This is a huge field which cannot be exhaustively covered. We give below a few examples of well-known side reactions. • Acetone in acidic media is easily transformed into mesityl oxide:

So do not forget to test for it when performing residual solvents analysis on drugs. • In basic medium the diketone-alcohol is obtained:

• DMF can be hydrolyzed in presence of hydrochloric acid:

• Acids undergoing reaction in alcoholic media can be partially transformed into esters

• Transesterification reaction. Take care when, for example, recrystallization has to be performed for a molecule containing an ester group:

• Aldehydes (even ketones) can be transformed in alcoholic solutions into ketals:

984


• From time to time the solvent can react in lieu of the reagent. In a synthesis aimed to prepare 3-chloro-1-methoxy-2-propan-2-ol starting from chloromethyloxirane and methanol, traces of ethanol present in methanol gave the corresponding ethoxylated compound:

Another and final example concerns the preparation of a urea derivative using the reaction of an amine and isocyanate in presence of isopropanol:

Other examples could be found. These obvious examples stress the need for close collaboration between chemists and analysts when elaborating chemical syntheses and corresponding quality control monographs. 14.21.1.2.2 Drug products11,12 14.21.1.2.2.1 General points Because ultimately it is the DP which is administered to the patient, it is necessary to have the quality of the solvents potentially used in the design of pharmaceutical formulations under control. 14.21.1.2.2.2 Areas of utilization Solvents including water are used in different ways in pharmaceutical formulation: • either as a part of the final drug product: injectables, drinkable solutions, patches, sprays, microemulsions • or used as an intermediary vehicle which is removed at the end of the process: granulation coating sugar coating microencapsulation


985

We have listed in Table 14.21.1.4. the most commonly used solvents. Table 14.21.1.4 Solvents used in formulation Water

Dichloromethane

Ethyl acetate

Chloroform

Ethyl alcohol (denaturated with butanol and isopropanol)

Hexane

Isopropyl alcohol (denaturated with methyl ethyl ketone)

Cyclohexane

Methanol

Polyethylene glycol (low molecular weight)

Acetone

Manufacturers try progressively to replace the formulations using organic solvents such as chloroform, dichloromethane, cyclohexane belonging to the class 2 (ICH classification see Chapter 16.2), for example by developing aqueous coatings. 14.21.1.2.2.3 What should be the quality? Taking into account the fact that the solvents used in the DP manufacturing process, either as a component of the formulation or as a residual solvent, will be absorbed by the patient, their quality must be of the highest standard. From a regulatory point of view, in almost every country if not all, it is mandatory to use solvents covered by a pharmacopoeial monograph (e.g., European Pharmacopoeia, USP, JP, local Pharmacopoeias). Some examples are given below. 14.21.1.3 Impacts of the nature of solvents and their quality on the physicochemical characteristics of raw materials and DP. 14.21.1.3.1 Raw materials (intermediates, DS, excipients) The impurities contained in the solvents could have several effects on the raw materials: • When the solvents are removed, non-volatile or less volatile impurities will be concentrated in raw materials. • They can induce chemical reactions leading to side products. • They can affect the stability of the raw material considered. • They can modify substantially the crystallization process. 14.21.1.3.1.1 Concentration of less volatile impurities Due to the potential concentration of these impurities, they should be tested for in both DS and excipients and even in intermediates of synthesis if the latter constitute the penultimate step of the synthesis and if solvents belong to class 1 or class 2 solvent (see Chapter 16.2). 14.21.1.3.1.2 Side reactions This case has already been illustrated (see paragraph 14.21.1.2.1.6.5). The skills of the chemist together with those of the analyst are needed to ensure that the presence of unexpected impurities can be detected. By way of example, the reactions involving the keto-enol tautomerism deserve to be mentioned. The equilibrium is very sensitive to the solvent so that the presence of other solvents as impurities in the main solvent can modify the keto-enol ratio leading to irreproducibility in the chemical process.13,14 14.21.1.3.1.3 Consequences for stability Some solvents, as mentioned in the paragraph 14.21.1.2.1.5, can contain very active entities such as aldehydes and peroxides. For example, if the raw material contains primary or sec-

986


ondary amines and/or is susceptible to oxidation or hydrolysis, it is likely that degradants will be formed over the time, reducing potentially the retest date of the raw material. The same situation could affect the DP and therefore two examples are given in paragraph 14.21.1.3.2. 14.21.1.3.1.4 Solvent purity and crystallization This issue is may be less known except for chemists working in this specialized area. Because the consequences can be important for the processability, stability and occasionally bioavailability of the drug substance in its formulation, it is relevant to comment on this subject. 14.21.1.3.1.4.1 Role of the nature and the quality of solvent on crystallization Most of the drugs on the market are obtained as a defined crystalline structure and formulated as solid dosage forms. It is well known that a molecule can crystallize to give different crystalline structures displaying what is called polymorphism. The crystal structures may be anhydrous or may contain a stoichiometric number of solvent molecules leading to the formation of solvates (hydrates in case of water molecules). Pseudopolymorphism is the term used to describe this phenomenon. Another characteristic which plays a major role in the overall processability of the DS for DP manufacture is the “crystal habit”. This term is used15,16 to describe the overall shape of crystals, in other words, the differing external appearance of solid particles which have the same internal crystalline structure. Both structures (internal, external) are under the control of different parameters including the nature of the solvent used and its quality. The role of the solvent itself in the overall crystallization process, including the determination of the crystal structure and the crystal habit is well known.17 But it is equally worth noting that impurities coming from: • the product to be crystallized • the solvent used • the environment can selectively affect the nucleation process and the growth rates of different crystal faces.17-21 They can be selectively adsorbed to certain faces of the polymorphs thereby inhibiting their nucleation or retarding their growth to the advantage of others. Crystal shape (habit) can also be modified by a solvent without polymorphic change. Additives or impurities can block, for a defined polymorph, the growth rate of certain faces leading e.g. to needles or plates. It is possible to introduce deliberately additives to “steer” the crystallization process. An interesting example of this crystal engineering strategy have been published for e.g., adipic acid22 or acetaminophen.23 14.21.1.3.1.4.2 Solvent-solid association/overview After the crystallization of the product, solvents must be removed in order to obtain the minimum amount of residual solvents compatible with safety considerations and/or physicochemical considerations including stability, processability and occasionally microbiological quality (see Chapter 16.2). Different situations can be encountered. 14.21.1.3.1.4.2.1 Solvent outside the crystal The solvent remains outside the crystals at the time of crystal formation. It is adsorbed on the surface or in the crystal planes. In the first case, the solvent is easily removed. But in the second case, if a cleavage plane exists, the drying process can be very difficult. Two methods can be used to try to remove this type of residual solvent almost completely.


987

• Displacement by water vapor in an oven, keeping in mind that this method may introduce some degradation leading to processability problems.11,24

• Extraction by supercritical CO225,26 but the extraction power of CO2 is basically limited to slightly polar solvents. 14.21.1.3.1.4.2.2 Solvent inside the crystal The solvent remains inside the crystalline structure. Three situations can arise. 14.21.1.3.1.4.2.2.1 Occluded solvents During rapid crystallization, some degree of disorder (amorphous phases, crystalline defects) can arise, creating pockets where residual solvent can be occluded. Through a process of dissolution/recrystallization this “hole” moves towards the external faces of the crystal releasing the solvent at the end. This phenomenon is more frequent for large crystals (500 µm/600 µm) but rare for smaller crystals (1 - 100 µm). The solvent odor which is detected when opening a drum or a bag containing a substance which was dried in the normal way can be explained by this mechanism. 14.21.1.3.1.4.2.2.2 Solvates At the end of the crystallization process, the substance can be isolated as a solvate (hydrate), i.e., as a pseudopolymorph. The solvates generally have quite different physicochemical properties from the anhydrous form. Their stability can be questionable and in any case deserves to be investigated. In some cases it is possible to remove the solvent from the crystal without changing the structure of the lattice leading to an isomorphic desolvate which displays a similar X-ray diffraction pattern to that of the parent compound.2,27 The lattice of the desolvated solvate is in a high energy state relative to the original solvate structure. A better dissolution rate and compressability can be expected,28 but the drawbacks are hygroscopicity and physico-chemical instability. The lattice could undergo a relaxation process over time which increases the packing efficiency of the substance by reducing the unit cell volume. When developing a new chemical entity all these aspects have to be considered to avoid unpleasant surprises during development or once the drug is on the market. Due to the need for process scale-up and of making the manufacturing process more industrial, changes are introduced especially in the crystallization and the drying processes, (e.g., change from static drying to dynamic drying). Because the drying is a particularly disturbing process for the integrity of the lattice, defects and/or amorphous phases may be created favoring subsequent polymorphic or pseudopolymorphic transformations of the crystalline form developed so far, if it is not the most stable one. 14.21.1.3.1.4.2.2.3 Clathrates In contrast to solvates, clathrates do not show any stoichiometric relationship between the number of molecules of the substance and the number of molecules of solvent. Clathrates actually correspond to a physical capture of solvent molecules inside the crystal lattice without any strong bonds including hydrogen bonds. Molecules of one or several solvents can be trapped within the crystalline structure as long as the crystallization has been performed with a pure solvent or a mixture. The case of the sodium salt of warfarin giving “mixed” clathrates with water and isopropyl alcohol is well known and the existence of 8/4/0 or 8/2/2 proportions has been shown.11 It is fairly obvious that some powder properties like wettability can be modified by the formation of clathrates. Because their formation is not easy to control, some batch to batch inconsistency may be expected in this situation.

988


Figure 14.21.1.1. DSC and TG patterns.

In order to demonstrate the internal character of the clathrate relative to the crystal lattice, the example of a molecule developed in the laboratory of one of the authors is given below. Figure 14.21.1.1 shows the DSC pattern of a molecule with the melting event at 254°C and the corresponding TG pattern obtained at the same temperature scanning rate. At the time the melting occurs, a loss of weight is observed corresponding to the loss of 0.2 % of isopropanol. The nature and the amount of the solvent have been confirmed by GC after dissolving the substance. 14.21.1.3.2 Drug product As for the DS, the solvents used for DP manufacture can produce some negative effects by themselves or through their own impurities. For liquid or semi liquid formulations, the formulator has to ensure that the solvents themselves do not display chemical interactions with the DS or the excipients. Everything which has been said in paragraphs 14.21.1.2.1.5 and 14.21.1.3.1 remains true here. 14.21.1.3.2.1 Interaction of impurities contained in the solvent As said in paragraph 14.21.1.3.1 with the DS, impurities contained in the solvent especially if they are strongly reactive, like aldehydes or peroxides, can promote formation of degradants. Regarding aldehydes, the publication of Bindra and all29 should be mentioned. It relates to the degradation of the o-benzylguanine in an aqueous solution containing polyethylene glycol 400 (PEG 400). This type of solvent very often contains formaldehyde, which can lead to the formation of a precipitate over time:


989

PEG can also contain peroxides which can initiate over time, the formation of degradants via an oxidation process. Several publications have dealt with this phenomenon.30,31 14.21.1.3.2.2 Interaction with the container When the formulation is a solution which is prepared from water and different organic solvents, it is mandatory to investigate possible interactions between the medium and the container especially if the latter is of polymeric nature (PVC-PVDC, polyethylene, etc.) with or without elastomeric stoppers. A thorough investigation is necessary including: • an examination of the solution for plasticizers, antioxidants, monomers and oligomers, mineral impurities, potentially extracted from the container, • the evaluation of the absorption by the container of components (DS, excipients) contained in the solution. In the first case, the migration of impurities into the solution could initiate physicochemical instability and possibly some potential toxicity. In the second case, a decrease in the content of the DS and/or some excipient (e.g., organic solvents added to promote the solubility) could lead to some loss of therapeutic efficacy and in some case to physical instability (precipitation). 14.21.1.3.2.3 Solvates formation during the solid dosage form manufacture During the granulation process it is possible that the DS (occasionally the excipient) could transform into a solvated crystalline structure (solvate, hydrate). During the drying process, different situations can occur: • The solvate is poorly stable and the solvent is easily removed leading to either the original polymorphic form but creating a certain degree of disorder in the crystalline structure or to what is called a “desolvate solvate” form. In this last case, also named “isomorphic desolvate”, the desolvated solvate retains the structure of its parent solvated form. The X-ray diffraction patterns look similar between the parent and the daughter forms. In this situation we have the creation of a molecular vacuum which could substantially impact on the stability, hygroscopicity and mechanical characteristics of the DS and finally of the DP. • The solvate is stable within the formulation: we then have in a sense a new chemical entity. The properties of the solvate could be entirely different (solubility, kinetics of dissolution, stability, processability, etc.) and the consequence could be either positive or negative. The case where the kinetics of dissolution are affected by the formation of solvates should always be investigated. Papers on this subject have been published for molecules such as lorazepam,32 hydrocortisone,33 cephalexin,32 etc. • Obviously, as in the case of raw materials (14.21.1.3.1.4.2.2.3) clathrate formation should be considered in order to explain possible batch to batch inconsistency. 14.21.1.3.3 Conclusions We have seen that the solvent, far from being inert, plays a key role by itself and occasionally via its own impurities in different ways which are important for pharmaceutical development. We will now discuss how to set up sound specifications for solvents in relation to their field of use.

990


14.21.1.4 Setting specifications for solvents 14.21.1.4.1 Solvents used for the raw material manufacture For the raw materials we should distinguish between solvents: • used during the synthesis • and those used for the last step of the manufacture corresponding very often to the crystallization process. As a rule of thumb, the specifications set Table 14.21.1.5. Examples of for solvents used for the crystallization step will monographs be more stringent than those used during the synthesis. Solvents used during the synthesis (of well For the intermediates of synthesis, if the controlled origin) origin of the solvent is under control (e.g., exisCharacter/Appearance tence of contracts/Quality Assurance audits) a ) Identification (IR, GC or n 20 simplified monograph is completely adequate D (see Table 14.21.1.5) as long as the supplier proPurity GC (generally not less than 98%) vides a detailed certificate of analysis where impurities (including solvents) are properly Table 14.21.1.6. Examples of specified with acceptable limits. If the same solmonographs vent is used for the crystallization step additional purity tests are necessary (Table 14.21.1.6). Final crystallization solvents (of well conFor economic reasons, it may be necessary trolled origin) to recycle solvents. If so, the containers should Character/Appearance be fully identified in terms of storage: Identification (IR, GC or n 20 • If solvents can be efficiently purified (e.g., by D) redistillation) they must comply with the Tests same specifications as those of fresh solvents Water content (0.1 to 0.5 depending on and consequently can be used in any solvent type) synthesis. Residue on evaporation: not more than • If they still contain volatile impurities 0.01 per cent resulting from the reaction they come from, Purity GC not less than 99 % (examine for they can be recycled only for this reaction. In denaturing agents and other potential imthis case, the impurities should be identified purities) and their possible impact on the reaction evaluated. In Tables 14.21.1.7 and 14.21.1.8 we have summarized possible specifications for a fresh batch of ethyl acetate used for a defined chemical reaction and those for the recycled solvent. We recommend working with reliable solvent suppliers who can give every assurance on the quality of solvents provided to avoid any “unpleasant surprises”. Water should be mentioned separately. If it is used during the synthesis of intermediates the quality “drinking water” can be used without any problems. But if water is used during the last step of the process, its quality must be in compliance with the requirements of purified water as they are described in several pharmacopoeias. In Table 14.21.1.9 the requirements for the Ph. Eur and USP are given as examples. Purified water is generally obtained from the drinking water. It undergoes demineralization by either distillation or an ion-exchange process. Particular attention has to be paid to microbiological quality.


991

Table 14.21.1.7. Example of monograph Table 14.21.1.8. Example of monograph applied to recycled solvent: Ethyl acetate applied to new solvent: Ethyl acetate

Specifications

Specifications Controls Characters

Clear liquid, colorless

Complies

1.370 to 1.373

or C - Gas chromatography

Gas chromatography

Complies

Water content

Not more than 2.0%

Related substances Complies

Assay Ethyl acetate (purity)

Clear liquid, colorless

Tests

or B - Refractive index

Characters

Standards

Identification

Identification A - Infrared spectrum

Controls

Standards

Not less than 99.5%

Methanol

Not more than 1.0%

Ethanol

Not more than 2.0%

Ethyl chloride

Not more than 2.0%

Others impurities (sum)

Not more than 2.0%

Assay Ethyl acetate (purity)

Not less than 99.5%

14.21.1.4.2 Solvents used for the DP manufacture There is no other choice than to use the quality of solvents defined by a Pharmacopoeia. It is true that there are still discrepancies between the pharmacopoeias of different countries. It is hoped that the ICH process dealing with the harmonization of quality, safety and efficacy amongst three main zones of the world (EU, USA, Japan) will progressively reduce the remaining differences in dossiers submitted to Regulatory Authorities and the way the data are evaluated. As examples, Tables 14.21.1.10, 14.21.1.11, and 14.21.1.12 summarize specifications for acetone, ethanol and isopropanol given by the Ph. Eur. and USP. As can be seen only the Ph. Eur. monograph makes reference to volatile impurities to be tested for by GC. 14.21.1.5 Quality of solvents and analysis The solvents, including water, are used in almost every area of analytical sciences: spectroscopy, chromatography, potentiometry, electrochemistry. They should be characterized by a set of properties making them suitable for use for their intended purpose. 14.21.1.5.1 Quality of solvents used in spectroscopy As a general requirement, the solvents used in spectroscopy should be transparent and stable towards the relevant range of wavelengths. They should be able to dissolve the substance to be examined and not contain impurities affecting the stability of the substance or the validity of the method (selectivity, repeatability, limit of detection, analytical response). Theoretically the solvent chosen should have minimal interaction with the solute. But what could be seen as a disadvantage could also be an important source of structural information. What is called the solvent effect can help in UV, IR and NMR spectroscopies34 e.g. in struc-

992


Table 14.21.1.9. Water quality

ture elucidation. In the book by Reichardt,1 data regarding the cut-off points of solvents commonly used in UV/visible spectroscopy are provided. The cut-off point is defined as the wavelength in the ultraviolet region at which the absorbance approaches 1.0 using a 1-cm cell path with water as the reference. In the same way the range of transparency for IR-solvents are given. Complete IR-spectra of organic solvents can be found in the “Stadler IR spectra handbook of common organic solvents” . The solvents suppliers usually provide catalogues including a “spectroscopic grade” allowing the user to make a sound choice. In the case of the NMR spectroscopy problems arise with the residual protonated part of deuterated solvents (1H-NMR) and the 13C-NMR absorption bands of compounds used as solvents. References can be found1 where detailed data are given regarding these points.


993

Table 14.21.1.10. Specification for acetone

A common problem which can be met regularly in chemistry is the identification of signals from common contaminants in solvents of medium quality (see paragraph 14.21.1.2.1.6.1). Gottlieb and Col35 have published data on NMR chemical shifts of trace impurities contained in common laboratory solvents, making NMR spectroscopy the instrument of choice as a tool for routine quality control. 14.21.1.5.2 Quality of solvents used in chromatography The aim of this paragraph is not to focus on strategies for solvent selection in order to achieve extraction or liquid chromatography. A detailed literature review has been published by Barwick36 on this matter allowing the user to design relevant methodology in any

994


Table 14.21.1.11. Specification for ethanol

Impurities as per European Pharmacopoeia, Supplement 2000: A - 1,1-diethoxyrethane (acetal); B - Acetaldehyde; C - Acetone; D - Benzene; E - Cyclohexane; F - Methanol; G Butan-2-one (methyl ethyl ketone); H - 4-methylpentan-2-one (methyl isobutyl ketone); I - Propanol; J Propan-2-ol; K - Butanol; L - Butan-2-ol; M - 2-methylpropanol (isobutanol); N - Furan-2-carbaldehyde (furfural); O - 2-methylpropan-2-ol (1,1-dimethyl alcohol); P - 2-methylbutan-2-ol; Q - Pentan-2-ol; R - Pentanol; S - Hexanol; T - Heptan-2-ol; U - Hexan-2-ol; V - Hexan-3-ol


995

Table 14.21.1.12. Specification for isopropanol

particular case. It is preferable to focus on the impact of impurities present in the solvent on the chromatographic performances. Impurities in and additives to solvents can cause several problems and artifacts in liquid and gas chromatography.37-39 Primarily they can be the origin of irreproducible separations, enhanced UV-background and even of mechanical problems. De Schutter and Col37 have investigated this problem in purity of solvents used in high-performance thin-layer chromatography. A way to improve the stability of the chromatographic system by minimizing the role of solvent impurities is to add deliberately a controlled amount of organic modifier. For example, Lauren and Col40 have applied this technique, using decanol for improving the stability of the LC used for analyzing carotenoids. Middleditch and Zlatkis41 have listed an impressive range of stabilizers and additives which can be found in solvents which may help the chromatographer in explaining the occurrence of artifacts in chromatography. Zelvensky and Col38 have determined by gas chromatography the most common impurities contained in solvents for liquid chromatography. The series of solvents investigated include acetonitrile, methanol, ethanol, dichloromethane, formic acid, dimethylformamide,

996


pyridine, tetrahydrofuran and dimethyl sulphoxide. Parsons and Col39 have performed a search for trace impurities in solvents commonly used for gas chromatographic analysis of environmental samples. 14.21.1.5.3 Quality of solvents used in titrimetry For titrimetric determinations performed in an aqueous medium it is highly recommended to use distilled water and to perform a blank determination if necessary. For determinations performed in non aqueous media, the solvents should be as anhydrous as possible and, of course, inert to the titrant and the substance. Their purity should be such that they do not contain impurities which could react with the substance to be analyzed. A blank titration should be performed if necessary. 14.21.1.6 Conclusions Far from being inert and not affecting the molecules dissolved in it, the solvent can affect the behavior of the solute in different ways. This chapter has aimed to support the idea that it is important for the chemist and the pharmacist to control the quality of the solvents used in the different areas of pharmaceutical activity. As we have tried to show, many pitfalls can be avoided during the development of a drug if a thorough investigation of the quality of the solvents used is carried out. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

C. Reichardt, Solvents and solvents effect in organic chemistry, 2nd edition, 1998, VCH Verlagsgesellschaft mbH. G.A. Stephenson, E.G. Groleau, R.L. Kleemann, W. Xu and D.R. Rigsbee, J. Pharm. Sci., 87, 536 (1998). A. Picot, Le bon usage des solvants, Information Toxicologique n° 3, Unité de prévention du risque chimique - CNRS (France) (1995). ICH Harmonised Tripartite Guideline, Impurities: Guideline for Residual Solvents. M. Gachon, STP Pharma Pratiques 1, 531 (1991). Carbomer Monograph - Eur. Ph. Addendum 2000, p 494. J.A. Riddik and W.B. Bungler, Organic Solvents, Wiley Intersciences, New-York, 1970, pp 552-571. G.P.J. Ravissot, STP Pharma Pratiques, 9, 3 (1999). See Reference 1 p 427. M. Debaert, STP Pharma Pratiques, 1, 253 (1991). A.M. Guyot-Hermann, STP Pharma Pratiques, 1, 258 (1991). D. Chulia, M. Deleuil, Y. Pourcelot, Powder Technology and Pharmaceutical Processes, Elsevier, Amsterdam, 1994. J.L. Burdett and M.T. Rodgers, J. Am. Chem. Soc., 86, 2105 (1964). M.T. Rodgers and J.L. Burdett, Can. J. Chem., 43, 1516 (1965). J. Haleblian and W. McCrone, J. Pharm. Sci., 58, 911 (1969). P. York, Int. J. Pharm., 14, 1 (1983). S. Khoshkhoo and J. Anwar, J. Phys. D.; Appl. Phys., 26, B90 (1993). J. Shyh-Ming, Diss. Abst. Int., 57 (10), 6402-B (1997). W. Beckmann and W.H. OTTO, Chem. E. Res. Des., 74, 750 (1996). N. Rodriguez-Hornedo and D. Murphy, J. Pharm. Sci., 88, 651 (1999). R. David and D. Giron in Powder Technology and Pharmaceutical Processes, Handbook of Powder Technology, Elsevier Sciences, Ed., Amsterdam 1994, pp 193-241. A.S. Myerson, S.M. Jang, J. Cryst. Growth, 156, 459 (1995). A.H.L. Chow, D.J.W. Grant, Int. J. Pharm., 42, 123 (1988). C. Lefebvre-Ringard, A.M. Guyot-Hermann, R. Bouché et J. Ringard, STP Pharma Pratiques, 6, 228 (1990). D.C. Messer, L.T. Taylor, W.N. Moore and W.E. Weiser, Ther. Drug Monit., 15, 581 (1993). M. Perrut, Information Chimie, 321, 166 (1990). R.R. Pfeiffer, K.S. Yang and M.A. Tucker, J. Pharm. Sci., 59, 1809 (1970). R. Hüttenrauch, Pharm. Ind., 45, 435 (1983). D.S. Bindra, T.D. William and V.J. Stella, Pharm. Res., 11, 1060 (1994).

14.21.2 Predicting cosolvency

30 31 32 33 34 35 36 37 38 39 40 41

997

J.W. McGinity and J.A. Hill, J. Pharm. Sci., 64, 356 (1975). D.M. Johnson and W.F. Taylor, J. Pharm. Sci., 73, 1414 (1984). J. Joachim, D.D. Opota, G. Joachim, J.P. Reynier, P. Monges and L. Maury, STP Pharma Sci., 5, 486 (1995). S.R. Byrn, C.T. Lin, P. Perrier, G.C. Clay and P.A. Sutton, J. Org. Chem., 47, 2978 (1982). J. Wiemann, Y. Pascal, J. Chuche, Relations entre la structure et les propriétés physiques, Masson edit, Paris, 1965. H.E. Gottlieb, V. Kotlyar and A. Nudelman, J. Org. Chem., 62, 7512 (1997). V.J. Barwick, Trends in An. Chem., 16, 293 (1997). J.A. de Schutter, G. Van der Weken, W. Van den Bossche and P. de Moerloose, Chromatographia, 20, 739 (1985). V.Y. Zelvensky, A.S. Lavrenova, S.I. Samolyuk, L.V. Borodai and G.A. Egorenko, J. Chromatogr., 364, 305 (1986). W.D. Bowers, M.L. Parsons, R.E. Clement, G.A. Eiceman and F.W. Rarasek, J. Chromatogr., 206, 279 (1981). D.R. Lauren, M.P. Agnew, D.E. McNaughton, J. Liq. Chromatogr., 9, 1997 (1986). B.S. Middleditch, A. Zlatkis, J. Chromatogr. Sci., 25, 547 (1987).

14.21.2 PREDICTING COSOLVENCY FOR PHARMACEUTICAL AND ENVIRONMENTAL APPLICATIONS

An Li School of Public Health, University of Illinois at Chicago, Chicago, IL, USA

14.21.2.1 Introduction Cosolvency refers to the effects of adding one or more solvents (cosolvents), which are different from the existing solvent in a solution, on the properties of the solution or behavior of the solute. Cosolvency has found its applications in numerous engineering and scientific disciplines. The discussion in this section will be limited to aqueous phase cosolvency (the primary solvent is water), and cosolvents will include only pure organic solvents which are miscible with water either completely (in any proportion) or partially (in only certain proportions). The extent of cosolvency will be quantitatively described by the difference in solute solubilities in pure water and in a mixture of water and cosolvent(s). Cosolvency has been studied for decades. However, it remains a poorly understood phenomenon due in large measure to our limited awareness of the liquid structure and the intermolecular forces. At present, practical approaches to predicting cosolvency are to develop models based on established theories and to make use of correlation between experimental observations and properties of the substances involved. As with all modeling efforts, it is essential to make judicious simplifications at various levels. The efforts to date have given rise to several models, including the extended regular solution theory1-3 and its modification,4 excess free energy model,5-7 the phenomenological model,8,9 modified Wilson model,10,11 the combined nearly ideal binary solvent (NIBS) model,12 the mixture response surface model,13 and others. Many of these models, however, are considered to be more descriptive than predictive, because they inevitably involve one or more model parameters which are usually specific to a particular solute/solvent/cosolvent(s) system, and must be estimated from experimental data of solubility obtained for that system. On the other hand, purely empirical models, e.g., the double-log exponential equation,14 aim at satisfying mathematical descriptions of measured data, and often offer little insight to the process. Comparisons among cosolvency models have been made in several published papers.15-19

998

An Li

This discussion is intended to provide an easy approach for predicting the effect of cosolvents, which are frequently involved in pharmaceutical and environmental applications, on the solubility of organic chemicals. The starting point is the widely used log-linear model. 14.21.2.2 Applications of cosolvency in pharmaceutical sciences and industry Although many drugs are formulated and administrated in solid, vapor, powder, or other forms, using solutions as drug delivery vehicle has significant advantages. Most parenterally administered medicines are in the form of liquid solution, and for intravenous injection, liquid form dosage is the only possible form. However, drugs are usually designed with little concern about their level of solubility in solution. It is the task of pharmaceutical formulators to find appropriate forms for the drug to be effectively delivered into biological systems. The phenomena of cosolvency have been studied for more than a century by pharmaceutical scientists. Numerous experimental data are published in the literature, and most of the models mentioned in the introduction section have been developed from pharmaceutical research. In addition to the need of solubilizing drugs which are poorly water-soluble, controlling the dissolution of drugs administered as solids to optimize therapeutic activity also demands an improved understanding of drug solubilization. Approaches that may be pursued to enhance drug solubility in a liquid dosage formulation include adjusting pH, adding surfactants, cosolvents, or complexation agents. Choice of these techniques depends primarily on the drug’s chemical structure and physicochemical properties. For example, control of pH is applicable only when the drug is an electrolyte. To solubilize nonelectrolyte drugs, the use of cosolvents outweighs surfactants and complexing agents.20 Cosolvents that are routinely used in drug formulation include ethanol, propylene glycol, polyethylene glycol, and glycerin. Examples of pharmaceutical products containing these cosolvents are summarized in Table 14.21.2.1. Table 14.21.2.1. Selected pharmaceutical products containing cosolvents Trade Name

Cosolvent

Aclovate cream

Propylene glycol

Alurate elixir

Ethanol

Amidate

Propylene glycol

Amphojel

Glycerin

Apresoline

Propylene glycol

Aristocort cream

Propylene glycol

Ativan

Polyethylene glycol Propylene glycol

Bentyl syrup

Propylene glycol

Brevibioc

Ethanol

Cleocin T lotion

Glycerin

vol%

Manufacturer

Type

Schering

Topical

20

Roche

Oral liquid

35

Abbott

Parenteral

Wyeth-Ayerst

Oral

Ciba

Parenteral

Fujisawa

Topical

Wyeth-Ayerst

Parenteral

Lakeside

Oral

DuPont

Parenteral

Upjohn

Topical

10

20 80

25


Trade Name

999

Cosolvent

Comtrex cough

Ethanol

Cyclocort lotion

vol% 20

Manufacturer

Type

Bristol

Oral liquid

Polyethylene glycol

Lederle

Topical

Delsym

Propylene glycol

McNeil Consumer

Oral

Depo-Medrol

Polyethylene glycol

Upjohn

Parenteral

Dilantin

Ethanol Propylene glycol

10 40

Parke-Davis

Parenteral

Dramamine

Propylene glycol

50

Searle

Parenteral

Elocon lotion

Propylene glycol

Schering

Topical

Entex liquid

Glycerin

Norwich-Eaton

Oral

Fluonid solution

Propylene glycol

Herbert

Topical

Halog cream

Propylene glycol

Westwood-Squibb

Topical

Halog ointment

Polyethylene glycol

Westwood-Squibb

Topical

Kwell cream

Glycerin

Reed & Carnrick

Topical

Lanoxin


10 40

Librium

Propylene glycol

20

Lidex

Propylene glycol

Luminal Sod

Propylene glycol

67.8

MVI-12

Propylene glycol

Nembutal


Neoloid

Propylene glycol

Nitro-BID IV

Ethanol

Novahistine DH

Glycerin

Paradione

Ethanol

Pentuss

Burroughs Wellcome

Parenteral

Roche

Parenteral

Syntex

Topical

Winthrop

Parenteral

30

Armour

Parenteral

10 40

Abbott

Parenteral

Lederle

Oral

Marion

Parenteral

Lakeside

Oral

Abbott

Oral liquid

Propylene glycol

Fisons

Oral

Psorcon ointment

Propylene glycol

Dermik

Topical

Rondec DM

Glycerin

Ross

Oral

S-T Forte syrup

Ethanol

Scot-Tussin

Oral liquid

Sulfoxyl lotion

Propylene glycol

Stiefel

Topical

Tinactin

Polyethylene glycol

Schering

Topical

Trideslon cream

Glycerin

Miles

Topical

Tussar

Propylene glycol

Rorer

Oral

Tussionex

Propylene glycol

Fisons

Oral

70

65

5

1000

An Li

Trade Name

Cosolvent

Tylenol

Propylene glycol

Valium


Vepesid

Ethanol

vol%

10 40 30.5

Manufacturer

Type

McNeil Consumer

Oral

Roche

Parenteral

Bristol-Myers

Parenteral

(Data are from reference 21)

14.21.2.3 Applications of cosolvency in environmental sciences and engineering The significance of cosolvency research in environmental sciences stems from the need for accurately modeling the distribution and movement of organic pollutants, and cleaning up polluted soils and sediments. Since the late 1970s, environmental research on the effect of cosolvents has grown steadily. Most published research papers have focused on the effects of adding cosolvents on the aqueous solubility18,22-31 and soil sorption28,32-44 of pollutants of interest. A few researchers have also examined cosolvent effects on liquid phase partitioning.45,46 In the cases of industrial waste discharges, liquid fuel and paint spills, storage tank leakage, landfill leaching, and illegal dumping, various organic solvents may find their way into the natural environment. These solvents may not only act as pollutants themselves, but also bring substantial changes on the distribution, movement, and fate of other environmental pollutants with high concern. In environmental cosolvency studies, the majority of the solutes are hydrophobic organic compounds (HOCs), including benzene and its derivatives, polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), polychlorinated dibenzo-p-dioxins and furans (PCDDs and PCDFs), and various pesticides. These chemicals are toxic, and many of them are mutagenic and carcinogenic. The 1990 Clean Air Act Amendment has stimulated research on gasoline additives such as methyl t-butyl ether (MTBE) and formulated fuels like gasohol. Their cosolvent effects on the solubility and sorption of a few pollutant groups have also been examined.25,47 Meanwhile, environmental engineers have put cosolvents to work in cleaning up contaminated sites. As a consequence of the failure of using traditional pump-and-treat remediation for soils contaminated with organic pollutants, a few new approaches have been experimented since the late 1980s. Among those involving cosolvents, Ex situ solvent extraction was developed to treat excavated soils, sediment, or sludge. A typical one is the basic extractive sludge treatment (B.E.S.T.) process certified by USEPA.48,49 Triethylamine was selected as the extracting solvent due mainly to its inverse miscibility property - it is completely miscible with water below 60oF but separates from water above 90oF. This property makes it easier to recycle the solvents after separating the treated solids from liquids containing the solvent, pollutants, and water. For PCBs in various soils, it is typical to achieve an extraction efficiency higher than 99% using the B.E.S.T technique. More attractive are in situ remediation approaches, which often cost less. Cosolvents promote the mobilization of organic chemicals in soils, thus accelerating the cleanup of contaminated site. Cosolvent flushing has been developed using the same principles as those used in solvent flooding, a technique to enhance petroleum recovery in oil fields. It involves injecting a solvent mixture, mostly water plus a miscible cosolvent, into the vadose


1001

or saturated zone upgradient of the contaminated area. The solvent with the removed contaminants is then extracted downgradient and treated above ground. Precise formulations for the water/cosolvent mixture need to be determined by laboratory and pilot studies in order to achieve the desired removal.50-53 A few field-scale evaluations of this technique were carried out at Hill Air Force Base, Utah, where the aquifer had been severely contaminated by jet fuel, chlorinated solvents, and pesticides during 1940s and 1950s. These contaminants had formed a complex non-aqueous phase liquid (NAPL) containing more than 200 constituents, which covered the surfaces of soil particles and was trapped in pores and capillaries over the years. One of the evaluations consisted of pumping ternary cosolvent mixture (70% ethanol, 12% n-pentanol, and 18% water) through a hydraulically isolated test cell over a period of 10 days, followed by flushing with water for another 20 days.54,55 The removal efficiency varied from 90-99% at the top zone to 70-80% at the bottom near a confining clay layer. Similar removal efficiencies were obtained from another test cell using a combination of cosolvent n-pentanol and a surfactant at a total of 5.5 wt % of the flushing solution.56 In order to remove gasoline residuals at a US Coast Guard base in Traverse City, Michigan, it was demonstrated that the contaminants were mobilized when cosolvent 2-propanol was used at 50% concentration, while methanol at either 20% or 50% showed little effect.57 Cosolvent flushing was also proven to be effective in treating NAPLs which were denser than water. Methanol, isopropanol, and t-butanol were used in treating soils contaminated with tri- and tetra-chlorinated ethylenes.58 The applicability of solvent flushing, however, is often limited by the characteristics of the soil, especially the particle size distribution. While sandy soils may result in uncontrolled fluid migration, clayey soils with particles size less than 60 µm are often considered unsuitable for in situ solvent flushing due to low soil permeability. In an attempt to remove PAHs from poorly permeable soils, Li, et al.59 investigated the possibility of combining cosolvent flushing with the electrokinetic technique. Electrokinetic remediation involves application of a low direct electrical current to electrodes that are inserted into the ground. As water is continuously replenished at anodes, dissolved contaminants are flushed toward the cathode due to electroosmosis, where they can be extracted and further treated by various conventional wastewater treatment methods. Their column experiment of removing phenanthrene from soil was moderately successful with the assistance of cosolvent n-butylamine at 20%(v). Retardation factor (ratio of the water linear velocity to that of the chemical) of phenanthrene was reduced from 753 in pure water to 11 by the presence of n-butylamine, and 43% of the phenanthrene was removed after 127 days or 9 pore volumes. However, significant removal of phenanthrene was not attained in their experiments with acetone and hydrofuran as cosolvents. 14.21.2.4 Experimental observations Numerous experimental data exist in the literature on the solubility of organic solutes, including both drugs and environmental pollutants, in various mixtures of water and cosolvents. Experimental observations are often illustrated by plotting the logarithm of solubility of the solute versus the volume fraction of cosolvent in the solvent mixture. A few examples of solubilization curves are shown in Figure 14.21.2.1, which shows three typical situations for solutes of different hydrophobicity in the mixture of water and ethanol. The classification of solute/cosolvent/water systems based on their relative polarity was suggested by Yalkowsky and Roseman.61 Solutes which are less polar than both water

1002

An Li

Figure 14.21.2.1. Effects of ethanol on the solubilities of selected organic compounds. (a): n benzene, o naphthalene, ¡ biphenyl, ¯ anthracene, p benzo(a)pyrene, r perylene, ¡ chrysene; (b) n hydantoic acid, o hydantoin, ¡ methyl hydantoic acic, ¯ 5-ethyl hydantoin, r 5-isobutyl hydantoin; (c) o triglycine, ¡ diglycine, r glycine [Adapted, by permission, from Li and Yalkovsky, J. Pharm. Sci., 83, 1735 (1994).]

and the cosolvent are considered as “nonpolar”, those which have a polarity between those of water and the cosolvent as “semipolar”, and those which are more polar than both water and cosolvent as “polar”. Figure 14.21.2.1-a illustrates the behavior of relatively hydrophobic compounds, which tend to have monatonically increasing solubilization curves. The solubility enhancement is greater for the more hydrophobic solutes. Curves with opposite trends were mostly observed for polar solutes. The monatonical desolubilization is greater for more hydrophilic solutes, as evidenced by the curves in Figure 14.21.2.1-c. In-between are semipolar solutes with slightly parabolic curves shown in 14.21.2.1-b. The impact of adding cosolvents is much less profound for the semipolars than for the other two groups. On a linear solubility scale, the parabola tends to be more obvious than on the log scale. The same general trends were seen for the cosolvents glycerine61 and propylene glycol61,62 and presumably many other water-miscible cosolvents. It is more difficult to evaluate the effects of cosolvents which have limited miscibility with water. In the literature, such organic solvents have been termed as both cosolvents and cosolutes, and there is no clear criteria for the distinction. Cosolvent is usually miscible with water, or to be used in an attempt to increase the aqueous solubility of the solute. Cosolute, on the other hand, may be organic chemicals which have a similar chemical structure or behave similarly with the solute when they exist in water alone. The effects of cosolutes have been examined in a limited number of published papers.63-73 Partially water-miscible organic solvents (PMOSs) may act as either cosolvents or cosolutes, and the research in the past has shown the complexity of their effects.23,27-30,73-75 It was demonstrated that in order to exert effects on solubility or sorption of HOCs, PMOSs must exist as a component of the solvent mixture in an appreciable amount: Munz and Roberts23 suggested a mole fraction of greater than 0.005 and Rao and coworkers27,28 proposed a volume percent of 1% or a concentration above 104 mg/L. Cosolvents with relatively high water solubility are likely to demonstrate observable effects on the solubilities of solutes, up to their solubility limits, in a similar manner to cosolvents of complete miscibility with water. A few experimental examples of the effects of PMOSs include 1-butanol and


1003

1-pentanol acting on PCB congeners30 and naphthalene.26 and butanone on anthracene and fluoranthene.75 Even more hydrophobic organic solvents produce little or even negative influence on the solubility of HOCs. For instance, the presence of benzene does not increase the aqueous solubility of PCBs up to their saturation concentration.29 Solubility of a few PCB congeners in water were found to be depressed by dissolved dichloromethane and chloroform.73 On the other hand, PCB solubility showed little change when cosolvent benzyl alcohol, 1-hexanol, 1-heptanol, or 1-octanol was present.29,30 Similar “no change” observations were made for naphthalene with cosolvents dichloromethane and chloroform,73 and for solutes benzene and hexane with cosolvent MTBE.25 Much of the complexity with hydrophobic cosolvents, or rather, cosolutes, can be explained by the fact that these cosolvents may partition into the solute phase, thus the physical state of the solute is no longer the same as is in pure water. Instead of a basically pure crystalline or liquid phase of solute, the solute and the cosolvent form an organic mixture, and the composition and ideality of this mixture will very much determine the concentrations of its components in the aqueous phase. Such a situation may be better investigated along the line of phase partitioning, where Raoult’s law defines an ideal system. 14.21.2.5 Predicting cosolvency in homogeneous liquid systems The log-linear model Yalkowsky and Roseman introduced the log-linear model in 1984 to describe the phenomenon of the exponential increase in aqueous solubility for nonpolar organic compounds as the cosolvent concentration is increased.61 They showed that log S mi = f log S c + (1 − f ) log Sw

[14.21.2.1]

Rearranging equation [14.21.2.1] results in

(

log S mi / Sw

) = f log(S

c

/ Sw ) = σf

[14.21.2.2]

The left side of equation [14.21.2.2] reflects the extent of solubilization; f defines how much cosolvent is required to reach the desired solubilization. The constant σ is the end-to-end slope of the solubilization curve and defined by: σ = log S c − log Sw = log(S c / Sw )

[14.21.2.3]

The model can be extended to systems containing a number of cosolvents:

(

log S mi / Sw

)=∑σ f

i i

[14.21.2.4]

where the subscribe i denotes the ith component of the solvent mixture. Two measured solubilities will define the value of σ that is specific to a solute/cosolvent pair. The value of σ is also dependent of the solubility unit selected and on whether 10-based or e-based logarithm is used. The magnitude of σ reflects the difference in molecular interactions between solute/cosolvent and solute/water. When applied to describe cosolvency, σ is like a microscopic partition coefficient if water and cosolvent are thought of as two independent entities. There had been other definitions of σ, such as the

1004

An Li

partial derivative ∂(log Sm)/∂f,76 or the regressional (not end-to-end) slope of the solubilization curve.77 The σ defined in these ways will depend on the range of f and on the accuracy of all data points over the entire range of f. These definitions are not desirable because they make σ difficult to predict and interpret in light of the concept of ideal solvent mixture on which the log-linear model is based. Note also that σ is not related to the crystalline structure of the solute, since the contributions from the free energy of melting to the two solubilities cancel out. However, it may change if the solute exists in pure cosolvent with a chemical identity different from that in water, as in the cases where solute degradation, solvation, or solvent-mediated polymorphic transitions occur in either solvent. Estimation of σ Laboratory measurements of Sw and Sc can be costly and difficult. Various methods, including group contribution technique and quantitative structure (or property) property relationships (QSPRs or QPPRs),78 are available to estimate Sw and Sc, from which σ values can be derived. A direct approach of predicting σ has also been established based on the dependence of cosolvency on solute hydrophobicity. Among a number of polarity indices, octanol/water partition coefficient, Kow, was initially chosen by Yalkowsky and Roseman61 for correlation with σ, due mainly to the abundance of available experimental Kow data and the wide acceptance of the Hansch-Leo fragment method79 for its estimation. Kow is a macroscopic property which does not necessarily correlate with micro-scale polarity indices such as dipole moment, and only in a rank order correlates with other macroscopic polarity indicators such as surface tension, dielectric constant, and solubility parameter. Correlation between σ and solute Kow takes the form: σ = a + b log K ow

[14.21.2.5]

where a and b are constants that are specific for the cosolvent but independent of solutes. Their values have been reported for various cosolvents and are summarized in Table 14.21.2.2. From Table 14.21.2.2, the slopes of equation [14.21.2.5], b, are generally close to unity, with few below 0.6 or above 1.2. Most of the intercepts a are less than one, with a few negative values. In searching for the physical implications of the regression constants a and b, Li and Yalkowsky80 derived equation [14.21.2.6]:

(

)

(

)

(

σ = log K ow + log γ ∞0 * / γ c + log γ w / γ w∞* + log V0* / Vc

)

[14.21.2.6]

According to this equation, σ~log Kow correlation will indeed have a slope of one and a ∞* predictable intercept of log (Vo*/Vc) if both γ ∞* 0 / γ c and γ w / γ w terms equal unity. Vo* = -1 0.119 L mol based on a solubility of water in octanol of 2.3 mol L-1, and Vc ranges from 0.04 to 0.10 L3 mol-1, thus log (Vo*/Vc) is in the range of 0.08 to 0.47, for the solvents included in Table 14.21.2.2 with the exclusion of PEG400. However, both γ ∞* 0 / γ c and γ w / γ w∞* are not likely to be unity, and their accurate values are difficult to estimate for many solutes. The ratio γ ∞* 0 / γ c compares the solute behavior in water-saturated octanol under dilute conditions and in pure cosolvent at saturation, while the γ w / γ w∞* term reflects both the effect of dissolved octanol on the aqueous activity coefficient and the variation of the activity coefficient with concentration. Furthermore, the magnitudes of both terms will vary from one solute to another, making it unlikely that a unique regression intercept will be ob-


1005

served over a wide range of solutes. Indeed, both a and b were found to be dependent on the range of the solute log Kow used in the regression. For instance, for solutes with log Kow ≤ 0, 0.01 to 2.99, and ≥3, the correlation of σ versus log Kow for cosolvent ethanol have slopes of 0.84, 0.79, and 0.69, respectively, and the corresponding intercepts increase accordingly; the slope of the overall correlation, however, is 0.95. Much of the scattering on the σ~log Kow regression resides on the region of relatively hydrophilic solutes. Most polar solutes dissociate to some extent in aqueous solutions, and their experimental log Kow values are less reliable. Even less certain is the extent of specific interactions between these polar solutes and the solvent components. According to equation [14.21.2.6], σmay not be a linear function of solute log Kow on a theoretical basis. However, despite the complexities caused by the activity coefficients, quality of the regression of σ against log Kow is generally high, as evidenced by the satisfactory R2 values in Table 14.21.2.2. This can be explained by the fact that changes in both γ ratios are much less significant compared with the variations of Kow for different solutes. In addition, the two γ terms may cancel each other to some degree for many solutes, further reducing their effects on the correlation between σ and log Kow. It is convenient and reliable to estimate σ from known log Kow of the solute of interest, especially when the log Kow of the solute of interest falls within the range used in obtaining the values of a and b. Dependence of σ on the properties of cosolvents has been less investigated than those of solutes. While hundreds of solutes are involved, only about a dozen organic solvents have been investigated for their cosolvency potentials. A few researchers examined the correlations between σ and physicochemical properties of cosolvents for specific solutes, for instance, Li et al. for naphthelene,31 and Rubino and Yalkowsky for drugs benzocaine, diazepam, and phenytoin.81 In both studies, hydrogen bond donor density (HBD), which is the volume normalized number of proton donor groups of a pure cosolvent, is best for comparing cosolvents and predicting σ. Second to HBD are the solubility parameter and interfacial tension (as well as log viscosity and ET-30 for naphthalene systems), while log Kow, dielectric constant, and surface tension, correlate poorly with σ. The HBD of a solvent can be readily calculated from the density and molecular mass with the knowledge of the chemical structure using equation [14.21.2.7]. The disadvantage of using HBD is that it cannot distinguish among aprotic solvents which have the same HBD value of zero. HBD = (number of proton donor groups)(density)/(molecular mass)

[14.21.2.7]

In an attempt to generalize over solutes, Li and Yalkowsky82 investigated the possible correlations between cosolvent properties and slope of the σ~log Kow regressions (b). Among the properties tested as a single regression variable, octanol-water partition coefficient, interfacial tension, and solubility parameter, are superior to others in correlating with b. Results of multiple linear regression show that the combination of log Kow and HDB of the cosolvent is best (equation [14.21.2.8]). Adding another variable such as solubility parameter does not improve the quality of regression. b = 0.2513 log Kow - 0.0054 HBD + 1.1645

[14.21.2.8]

(N = 13, R2 = 0.942, SE = 0.060, F = 81.65) where log Kow (range: -7.6 ~ 0.29) and HBD (range: 0 ~ 41) are those of the cosolvent.

1006

An Li

Equation [14.21.2.8] can be helpful to obtain the b values for cosolvents not listed in Table 14.21.2.2. In order to estimate cosolvency for such cosolvents, values of the intercept a are also needed. However, values of a can be found in Table 14.21.2.2 for only about a dozen cosolvents, and there is no reliable method for its estimation. To obtain a for a cosolvent, a reasonable starting value can be log (Vo*/Vc), or -(0.92 + log Vc). The average absolute difference between a values listed in Table 14.21.2.2 and log (Vo*/Vc) is 0.18 (N=7) for alcohols and glycols, 0.59 (N = 6) for aprotic cosolvents, and > 1 for n-butylamine and PEG400.82 Table 14.21.2.2. Summary of regression results for relationship between σ and solute log Kow Cosolvent

N

a

b

R2

log Kow range

Ref.

Methanol

79

0.36±0.07

0.89±0.02

0.96

-4.53 ~ 7.31

80

Methanol

16

1.09

0.57

0.83

2.0 ~ 7.2

29

Methanol

16

1.07

0.68

0.84

n.a.

24

Ethanol

197

0.30±0.04

0.95±0.02

0.95

-4.90 ~ 8.23

80

Ethanol

107

0.40±0.06

0.90±0.02

0.96

-4.9 ~ 6.1

60

Ethanol

11

0.81

0.85

0.94

n.a.

24

1-Propanol

17

0.01±0.13

1.09±0.05

0.97

-3.73 ~ 7.31

80

2-Propanol

20

-0.50±0.18

1.11±0.07

0.94

-3.73 ~ 4.49

80

2-Propanol

9

0.63

0.89

0.85

n.a.

24

Acetone

22

-0.10±0.24

1.14±0.07

0.92

-1.38 ~ 5.66

80

Acetone

14

0.48

1.00

0.93

0.6 ~ 5.6*

24

Acetonitrile

10

-0.49±0.42

1.16±0.16

0.86

-0.06 ~ 4.49

80

Acetonitrile

8

0.35

1.03

0.90

n.a.

24

Dioxane

23

0.40±0.16

1.08±0.07

0.91

-4.90 ~ 4.49

80

Dimethylacetamide

11

0.75±0.30

0.96±0.12

0.87

0.66 ~ 4.49

80

Dimethylacetamide

7

0.89

0.86

0.95

n.a.

24

Dimethylformamide

11

0.92±0.41

0.83±0.17

0.73

0.66 ~ 3.32

80

Dimethylformamide

7

0.87

0.87

0.94

n.a.

24

Dimethylsulfoxide

12

0.95±0.43

0.79±0.17

0.68

0.66 ~ 4.49

80

Dimethylsulfoxide

7

0.89

0.87

0.95

n.a.

24

Glycerol

21

0.28±0.15

0.35±0.05

0.72

-3.28 ~ 4.75

80

Ethylene glycol

13

0.37±0.13

0.68±0.05

0.95

-3.73 ~ 4.04

80

Ethylene glycol

7

1.04

0.36

0.75

n.a.

24

Propylene glycol

62

0.37±0.11

0.78±0.04

0.89

-7.91 ~ 7.21

80


Cosolvent

1007

N

a

b

R2

log Kow range

Ref.

Propylene glycol

47

0.03

0.89

0.99

-5 ~ 7

61

Propylene glycol

8

0.77

0.62

0.96

n.a.

24

PEG400

10

0.68±0.43

0.88±0.16

0.79

-0.10 ~ 4.18

80

Butylamine

4

1.86±0.30

0.64±0.10

0.96

-1.69 ~ 4.49

80

*estimated from Figure 3 in Reference 24. n.a. = not available.

This empirical approach using equations [14.21.2.8], [14.21.2.5], and [14.21.2.2] can produce acceptable estimates of log (Sm/Sw) only if the solubilization exhibits a roughly log-linear pattern, such as in some HOC/water/methanol systems. In addition, it is important to limit the use of equations [14.21.2.5] and [14.21.2.8] within the ranges of log Kow used in obtaining the corresponding parameters. 14.21.2.6 Predicting cosolvency in non-ideal liquid mixtures Deviations from the log-linear model Most solubilization curves, as shown in Figure 14.21.2.1, exhibit significant curvatures which are not accounted for by the log-linear model. A closer look at the solubilization curves in Figure 14.21.2.1 reveals that the deviation can be concave, sigmoidal, or convex. In many cases, especially with amphiprotic cosolvents, a negative deviation from the end-to-end log-linear line is often observed at low cosolvent concentrations, followed by a more significant positive deviation as cosolvent fraction increases. The extent of the deviation from the log-linear pattern, or the excess solubility, is measured by the difference between the measured and the log-linearly predicted log Sm values:

(

)

log S m / S mi = log S m − (log Sw + ∑ σ i f i )

[14.21.2.9]

The values of log (Sm/Sim) for naphthalene, benzocaine, and benzoic acid in selected binary solvent mixtures are presented in Figures 14.21.2.2-a, -b, and -c, respectively. The log-linear model is based on the presumed ideality of the mixtures of water and cosolvent. The log-linear relationship between log (Sm/Sw) and f is exact only if the cosolvent is identical to water, which cannot be the case in reality. Deviation is fortified as any degradation, solvation, dissociation, or solvent mediated polymorphic transitions of the solute occur. The problem is further compounded if the solute dissolves in an amount large enough to exert significant influence on the activity of solvent components. Due to the complexity of the problem, efforts to quantitatively describe the deviations have achieved only limited success. A generally accepted viewpoint is that the deviation from the log-linear solubilization is mainly caused by the non-ideality of the solvent mixture. This is supported by the similarities in the patterns of observed log Sm and activities of the cosolvent in solvent mixture, when they are graphically presented as functions of f. Based on the supposition that solvent non-ideality is the primary cause for the deviation, Rubino and Yalkowsky87 examined the correlations between the extent of deviation and various physical properties of solvent mix-

1008

An Li

Figure 14.21.2.2a. Deviations from log-linear model (equation [14.21.2.2], triangle) and the extended log-linear model (equation [14.21.2.10], circle) for solute naphthalene in various water cosolvent systems. Experimental data are from Ref. 83.

tures. However, none of the properties consistently predicted the extrema of the deviation, although density corresponded in several cases. Non-ideality of a mixture is quantitatively measured by the excess free energy of mixing. From this standpoint, Pinal et al.75 proposed that a term Σ(fi ln γi) be added to equation [14.21.2.4] to account for the effect of the non-ideality of solvent mixture:

(

log S mii / Sw

)=∑σ f

i i

+ 2.303∑ f i log γ i

[14.21.2.10]

where γi is the activity coefficient of solvent component i in solute-free solvent mixture. Values of γ‘s can be calculated by UNIFAC, a group contribution method for the prediction of activity coefficients in nonelectrolyte, nonpolymeric liquid mixtures.88 UNIFAC derived activity coefficients are listed in Table 14.21.2.3 for selected cosolvent-water mixtures. They are calculated with UNIFAC group interaction parameters derived from vapor-liquid equilibrium data.89,90 The difference between the experimental log Sm and that predicted by the extended log-linear model, i.e., equation [14.21.2.10], is


1009

Figure 14.21.2.2b. Deviations from log-linear model (equation [14.21.2.2], triangle) and the extended log-linear model (equation [14.21.2.10], circle) for solute benzocaine in various water cosolvent systems. Experimental data are from Refs. 84 and 85.

(

)

log S m / S mii = log S m − (log Sw + ∑ σ i f i + 2.303∑ f i log γ i ) [14.21.2.11] Results of equation [14.21.2.11] for naphthalene, benzocaine, and benzoic acid in selected binary solvent mixtures are also included in Figure 14.21.2.2. A few other examples can be found in Pinal et al.75 The extended log-linear model outperforms the log-linear model in more than half of the cases tested for the three solutes in Figure 14.21.2.2. The improvement occurs mostly in regions with relatively high f values. In the low f regions, negative deviations of solubilities from the log-linear pattern are often observed as discussed above, but are not accounted for by the extended log-linear model as presented by equation [14.21.2.10]. In some cases, such as naphthalene in methanol and propylene glycol, and benzoic acid in ethylene glycol, the negative deviations occur over the entire f range of 0~1. In these cases, the extended log-linear model does not offer better estimates than the original log-linear model. With the activity coefficients listed in Table 14.21.2.3, the extended log-linear model generates worse estimates of log (Sm/Sw) than the log-linear model for systems containing dimethylacetamide, dimethylsulfoxide, or dimethylformamide. There is a possibility that

1010

An Li

Figure 14.21.2.2c. Deviations from log-linear model (equation [14.21.2.2], triangle) and the extended log-linear model (equation [14.21.2.10], circle) for solute benzoic acid in various water cosolvent systems. Experimental data are from Refs. 61 and 86.

the UNIFAC group interaction parameters involved in these systems are incorrect. With all the systems tested in this study with solute naphthalene, benzocaine, or benzoic acid, it is also found that replacing fi in the last term of equation [14.21.2.11] with mole fraction xi offers slight improvement in only a few cases. Dropping the logarithm conversion constant 2.303 results in larger estimation errors for most systems. An apparent limitation of this modification is the exclusion of any active role the solute may play on the observed deviation. Little understanding of the influence of solute structure and properties on deviations from the log-linear equation has been obtained. Although the patterns of deviations tend to be similar among solutes, as mentioned above, the extent of deviation is solute-dependent. For instance, C1~C4 alkyl esters of p-hydroxybenzoates and p-aminobenzoates demonstrated similar characteristics of solubilization by propylene glycol, with a negative deviation from the log-linear pattern occurring when f is low, followed by a positive one when f increases.91 The magnitude of the negative deviation, however, was found to be related to the length of the solute alkyl chain in each group, while that of the positive deviation to the type of the polar groups attached.91


1011

Both the hydrophobicity and hydrogen bonding property of the solutes seem to be important in influencing the extent of the deviation from the ideal log-linear pattern. Additional deviations related to the solute’s behavior may occur. For organic electrolytes, the acid dissociation constant Ka may decrease as cosolvent fraction f increases.40,75,92 This, in turn, will affect the patterns of solubilization by cosolvents. Furthermore, a high concentration of solutes may invalidate the log-linear model, which presumes negligible volume fraction of solute and no solute-solute interactions. For solid solutes, solvent induced polymorphism may also bring additional changes in their solubilization profile. Another approach to quantitatively address the deviations of solubilization from the log-linear model makes use of an empirical parameter β:

(

β = log(S m / Sw ) / log S mi / Sw

)

[14.21.2.12]

The modified log-linear equation then takes the form: log(S m / Sw ) = β∑ σ i f i Table 14.21.2.3. UNIFAC water-cosolvent systems f

0.1

Methanol

[14.21.2.13]

derived

0.2

activity

0.4

MW = 32.04

coefficients

0.6

for

selected

0.8

binary

0.9

Density = 0.7914

mol/L

2.4700

4.9401

9.8801

14.8202

19.7603

22.2303

x

0.0471

0.1000

0.2286

0.4001

0.6401

0.8001

γ, cosolvent

1.972

1.748

1.413

1.189

1.052

1.014

γ, water

1.003

1.013

1.055

1.14

1.298

1.424

Ethanol

MW = 46.07

Density = 0.7893

mol/L

1.7133

3.4265

6.8530

10.2796

13.7061

15.4194

x

0.0331

0.0716

0.1705

0.3163

0.5523

0.7351

γ, cosolvent

5.550

4.119

2.416

1.564

1.152

1.050

γ, water

1.005

1.022

1.097

1.256

1.57

1.854

1-Propanol

MW = 60.1

Density = 0.8053

mol/L

1.3399

2.6799

5.3597

8.0396

10.7195

12.0594

x

0.0261

0.0569

0.1385

0.2657

0.4910

0.6846

γ, cosolvent

12.77

8.323

3.827

2.001

1.248

1.077

γ, water

1.006

1.024

1.111

1.301

1.706

2.093

2-Propanol mol/L

MW = 60.1 1.3058

2.6116

Density = 0.7848 5.2233

7.8349

10.4466

11.7524

1012

An Li

f

0.1

0.2

0.4

0.6

0.8

0.9

x

0.0255

0.0555

0.1355

0.2607

0.4846

0.6790

γ, cosolvent

12.93

8.488

3.921

2.040

1.258

1.080

γ, water

1.006

1.023

1.107

1.294

1.695

2.084

Acetone

MW = 58.08

Density = 0.7899

mol/L

1.3600

2.7200

5.4401

8.1601

10.8802

12.2402

x

0.0265

0.0577

0.1403

0.2686

0.4947

0.6878

γ, cosolvent

8.786

6.724

3.952

2.370

1.484

1.196

γ, water

1.004

1.015

1.075

1.222

1.616

2.211

Acetonitrile

MW = 41.05

Density = 0.7857

mol/L

1.9140

3.8280

7.6560

11.4840

15.3121

17.2261

x

0.0369

0.0793

0.1868

0.3407

0.5795

0.7561

γ, cosolvent

10.26

7.906

4.550

2.536

1.501

1.126

γ, water

1.005

1.021

1.11

1.366

2.076

3.571

Dioxane

MW = 88.11

Density = 1.0329

mol/L

1.1723

2.3446

4.6891

7.0337

9.3783

10.5506

x

0.0229

0.0501

0.1233

0.2404

0.4577

0.6551

γ, cosolvent

14.71

8.743

3.616

1.833

1.171

1.124

γ, water

1.006

1.026

1.112

1.284

1.604

1.666

DMA

MW = 87.12

Density = 0.9429

mol/L

1.0823

2.1646

4.3292

6.4938

8.6584

9.7407

x

0.0212

0.0464

0.1149

0.2261

0.4380

0.6368

γ, cosolvent

0.121

0.141

0.204

0.330

0.602

0.826

γ, water

0.999

0.994

0.962

0.872

0.651

0.453

DMF

MW = 73.1

Density = 0.9445

mol/L

1.2921

2.5841

5.1683

7.7524

10.3365

11.6286

x

0.0252

0.0549

0.1342

0.2586

0.4819

0.6767

γ, cosolvent

0.833

0.873

0.930

0.962

0.983

0.985

γ, water

0.999

0.997

0.991

0.984

0.972

0.969

DMSO

MW = 78.13

Density = 1.10

mol/L

1.4079

2.8158

5.6316

8.4475

11.2633

12.6712

x

0.0274

0.0596

0.1445

0.2754

0.5034

0.6952

γ, cosolvent

0.07956

0.110

0.211

0.399

0.715

0.899

0.996

0.981

0.913

0.774

0.540

0.386

γ, water


f

0.1

Glycerol

1013

0.2

0.4

MW = 92.1

0.6

0.8

0.9

Density = 1.2611

mol/L

1.3693

2.7385

5.4771

8.2156

10.9542

12.3235

x

0.0267

0.0580

0.1411

0.2699

0.4964

0.6893

γ, cosolvent

1.257

1.066

0.903

0.899

0.969

0.996

γ, water

1.003

1.010

1.027

1.025

0.979

0.942

Ethylene glycol

MW = 62.07

Density = 1.1088

mol/L

1.7864

3.5727

7.1455

10.7182

14.2910

16.0773

x

0.0345

0.0744

0.1765

0.3254

0.5626

0.7432

γ, cosolvent

2.208

1.923

1.494

1.214

1.053

1.013

γ, water

1.002

1.01

1.047

1.12

1.247

1.338

Propylene glycol

MW = 76.09

Density = 1.0361

mol/L

1.3617

2.7234

5.4467

8.1701

10.8934

12.2551

x

0.0265

0.0577

0.1405

0.2688

0.4951

0.6881

γ, cosolvent

3.392

2.498

1.567

1.177

1.044

1.019

γ, water

1.005

1.018

1.069

1.145

1.224

1.267

Butylamine

MW = 73.14

Density = 0.7414

mol/L

1.0137

2.0273

4.0547

6.0820

8.1094

9.1231

x

0.0199

0.0436

0.1084

0.2149

0.4219

0.6215

γ, cosolvent

6.532

4.498

2.318

1.391

1.042

0.998

γ, water

1.004

1.016

1.071

1.175

1.326

1.384

Under the assumptions that the solute is chemically stable and has little influence on the activity of solvent component, β reflects the extent of deviation caused by the nonideality of the solvent mixture, as suggested by Rao et al.28 However, since β itself is a complicated function of f, equation [14.21.2.13] does not provide additional aid for predicting cosolvency. 14.21.2.7 Summary Applications of cosolvency in pharmaceutical and environmental research and industries are briefly summarized. Using ethanol as an example, the effects of adding a cosolvent on the solubilities of various organic solutes are presented in Figure 14.21.2.1. The log-linear solubilization model, equation [14.21.2.2] or [14.21.2.4], is the simplest theory of cosolvency developed so far. It discovers general trends and major determinant factors of cosolvency, thus providing guidelines for predicting solubility of organic chemicals in mixed solvents. The cosolvency power of a specific cosolvent towards a solute of interest, σ, can be estimated with equation [14.21.2.5] with the knowledge of the solute octanol-water partition coefficient Kow. Sources of error associated with this estimation method are discussed based on equation [14.21.2.6]. The slope of the σ~log Kow regression, b, can be

1014

An Li

estimated from the log Kow and hydrogen bond donor density of the cosolvent, as presented by equation [14.21.2.8]. One of the previously published modifications to the log-linear model, equation [14.21.2.10], is evaluated. The difference between the measured log Sm and those predicted by the log-linear and the extended log-linear model are presented in Figure 14.21.2.2 for solutes naphthalene, benzocaine, and benzoic acids in selected water and cosolvent mixtures. Notations a b f Kow Sc Sm Sim Siim Sw Vo* Vc Vw* Vw β σ γ ∞* 0 γc γw γ ∞w *

intercept of σ ~ log Kow regression slope of σ ~ log Kow regression volume fraction of cosolvent in mixed solvent with water. n-octanol water partition coefficient solubility in pure cosolvent solubility in the mixture of water and cosolvent solubility in the mixture of water and cosolvent, predicted by the log-linear model (Eq. [14.21.2.2]) solubility in the mixture of water and cosolvent, predicted by the extended log-linear model (Eq. [14.21.2.10]) solubility in pure water molar volume of 1-octanol saturated with water, 0.119 L mol-1 (based on a solubility of water in octanol of 2.3 mol L-1) molar volume of cosolvent molar volume of water saturated with 1-octanol, ≈ 0.018 L mol-1 molar volume of water, 0.018 L mol-1 empirically obtained water-cosolvent interaction parameter cosolvency power, σ = log (Sc/Sw) infinite dilution activity coefficient of solute in 1-octanol saturated with water activity coefficient of solute in cosolvent activity coefficient of solute in water infinite dilution activity coefficient of solute in water saturated with 1-octanol

References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

A. Martin, J. Newburger, and A. Adjel, J. Pharm. Sci., 68, 4 (1979). A. Martin, J. Newburger, and A. Adjel, J. Pharm. Sci., 69, 487 (1980). A. Martin, A. N. Paruta, and A. Adjel, J. Pharm. Sci., 70, 1115 (1981). P. Bustamante, B. Escalera, A. Martin, and E. Selles, Pharm. Pharmacol., 45, 253 (1993). N. A. Williams, and G. L. Amidon, J. Pharm. Sci., 73, 9 (1984). N. A. Williams, and G. L. Amidon, J. Pharm. Sci., 73, 14 (1984). N. A. Williams, and G. L. Amidon, J. Pharm. Sci., 73, 18 (1984). D. Khossrani, and K. A. Connors, J. Pharm. Sci., 82, 817 (1993). D. Khossravi, and K. A. Connors, J. Pharm. Sci., 81, 371 (1992). A. Jouyban-Gharamaleki, Chem. Pharm. Bull., 46, 1058 (1998). W. E. Acree, Jr., J. W. McCargar, A. I. Zvaigzne, and I. L. Teng, Phys. Chem. Liq., 23, 27 (1991) W. E. Acree, Jr. and A. I. Zvaigzne, Thermochimica. Acta, 178, 151 (1991). A. B. Ochsner, R. J. Belloto Jr., and T. D. Sololoski, J. Pharm. Sci., 74, 132 (1985). M. Barzegar-Jalali, and J. Hanaee, Int. J. Pharm., 109, 291 (1994). A. Li, and A. W. Andren, Environ. Sci. Technol., 29, 3001 (1995). J. K. Fu, and R. G. Luthy, J. Environ. Eng., 112, 328 (1986). M. Barzegar-Jalali, and A. Jouyban-Gharamaleki, Int. J. Pharm., 140, 237 (1996). R. M. Dickhut, D. E. Armstrong, and A. W. Andren, Environ. Toxicol. Chem., 10, 881 (1991). A. Jouyban-Gharamaleki, L. Valaee, M. Barzegar-Jalali, B. J. Clark, and W. E. Acree, Jr., Intern. J. Pharm., 177, 93 (1999). S. H. Yalkowsky, in Techniques of solubilization of drugs; S. H. Yalkowsky, Ed.; Dekker, New York, 1984, Chapter 1.


21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71

1015

S. C. Smolinske, Handbook of Food, Drug, and Cosmetic Excipients, CRC Press, Ann Arbor, MI, 1992. F. Herzel, and A. S. Murty, Bull. Environ. Toxicol., 32, 53 (1984). C. Munz, and P. Roberts, Environ. Sci. Technol., 20, 830 (1986). K. R. Morris, R. Abramowitz, R. Pinal, P. Davis, and S. H. Yalkowsky, Chemosphere, 17, 285 (1988). F. R. Groves, Jr, Environ. Sci. Technol., 22, 282 (1988). R. M. Dickhut, A. W. Andren, and D. E. Armstrong, J. Chem. Eng. Data, 34, 438 (1989). R. Pinal, P. S. C. Rao, L. S. Lee, P. V. Cline, and S. H. Yalkowsky, Environ. Sci. Technol., 24, 639 (1990). P. S. C. Rao, L. S. Lee, and A. L. Wood, EPA/600/M-91/009 (1991). A. Li, W. J. Doucette, and A. W. Andren, Chemosphere, 24, 1347 (1992). A. Li, and A. W. Andren, Environ. Sci. Technol., 28, 47 (1994). A. Li, A. W. Andren, and S. H. Yalkowsky, Environ. Toxicol. Chem., 15, 2233 (1996). P. Nkedi-Kizza, P. S. C. Rao, and A. G. Hornsby, Environ. Sci. Technol., 19, 975 (1985). J. K. Fu, and R. G. Luthy, J. Environ. Eng., 112, 346 (1986). R. W. Walters, and A. Guiseppl-Elie, Environ. Sci. Technol., 22, 819 (1988). P. Nkedi-Kizza, M. L. Brusseau, P. S. C. Rao, and A. G. Hornsby, Environ. Sci. Technol., 23, 814 (1989). A. L. Wood, D. C. Bouchard, M. L. Brusseau, and P. S. C. Rao, Chemosphere, 21, 575 (1990). P. S. C. Rao, L. S. Lee, and R. Pinal, Environ. Sci. Technol., 24, 647 (1990). F. C. Spurlock, and J. W. Biggar, Environ. Sci. Technol., 28, 1003 (1994). W. J. M. Hegeman, C. H. Van der Weijden, and J. P. G. Loch, Environ. Sci. Technol., 29, 363 (1995). L. S. Lee, and P. S. C. Rao, Environ. Sci. Technol., 30, 1533 (1996). V. A. Nzengung, E. A. Voudrias, P. Nkedi-Kizza, J. M. Wampler, and C. E. Weaver, Environ. Sci. Technol., 30, 89 (1996). R. P. Singh, Colloids and Surfaces A., 122, 63 (1997). V. A. Nzengung, P. Nkedi-Kizza, R. E. Jessup, and E. A. Voudrias, Environ. Sci. Technol., 31, 1470 (1997). T. C. Harmon, T. J. Kim, B. K. D. Barre, and C. V. Chrysikopoulos, J. Environ. Eng., January, 87 (1999). M. A. Ei-Zoobi, G. E. Ruch, and F. R. Groves Jr., Environ. Sci. Technol., 24, 1332 (1990). W. F. Lane, and R. C. Loehr, Environ. Sci. Technol., 26, 983 (1992). C. S. Chen, and J. J. Delfino, J. Environ. Eng., April 354 (1997). G. R. Jones, Environ. Prog., 11, 223 (1992). USEPA, Assessment and Remediation of contaminated Sediments (ARCS) Program: Remediation Guidance Document. EPA 905-R94-003, (1994) p.180. USEPA, EPA 542-K-94-006 (1995). C. T. Jafvert, Ground-Water Remediation Technologies Analysis Center, Echnology Evaluation Report, TE-96-02 (1996). R. W. Falta, GWMR, Summer, 94 (1998). D. C. M. Augustijin, R. E. Jessup, P. S. Rao, and A. L. Wood, J. Environ. Eng., 120, 42 (1994). R. K. Sillan, M. D. Annable, P. S. C. Rao, D. Dai, K. Hatfield, W. D. Graham, A. L. Wood, and C. G. Enfield, Water Resources Res., 34, 2191 (1998). P. S. C. Rao, M. D. Annable, R. K. Sillan, D. Dai, K. Hatfield, and W. D. Graham, Water Resources Res., 33, 2673 (1997). J. W. Jawitz, M. D. Annable, P. S. C. Rao, and R. D. Rhue, Environ. Sci. Technol., 32, 523 (1998). A. T. Kan, M. B. Tomson, and T. A. McRae, Proceedings of the 203rd American Chemical Society National Meeting, San Francisco, CA (1992). D. Brandes, and K. J. Farley, J. Water Environ. Res., 65, 869 (1993). A. Li, K. A. Cheung, and K. Reddy, J. Environ. Eng., 126, 527 (2000). A. Li, and S. H. Yalksowsky, J. Pharm. Sci., 83, 1735 (1994). S. H. Yalkowsky, and T. J. Roseman, in Techniques of solubilization of drugs; S. H. Yalkowsky, Ed.; Dekker, New York, 1984, Chapter 3. J. T. Rubino, and S. H. Yalkowsky, J. Pharm. Sci., 74, 416 (1985). P. J. Leinonen, and D. Mackay, Can. J. Chem. Eng., 51, 230 (1973). R. P. Eganhouse, and J. A. Calder, Geochimca et Cosmochimica Acta, 40, 555 (1976). Y. B. Tewari, D. E. Martire, S. P.; Wasik, and M. M. Miller, J. Solution Chem., 11, 435 (1982). S. Banerjee, Environ. Sci. Technol., 18, 587 (1984). D. R. Burris, and W. G. MacIntyre, Environ. Toxicol. Chem., 4, 371 (1985). H. H. Hooper, S. Michel, and J. M. Prausnitz, J. Chem. Eng. Data, 33, 502 (1988). D. Mackay, J. Contam. Hydrol. 8, 23 (1991). A. Li, and W. J. Doucette, Environ. Toxicol. Chem., 12, 2031 (1993). S. Lesage, and S. Brown, J. Contam. Hydrol., 15, 57 (1994).

1016

72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92

George Wypych

K. Brololm, and S. Feenstra, Environ. Toxicol. Chem., 14, 9 (1995). G. T. Coyle, T. C. Harmon, and I. H. Suffet, Environ. Sci. Technol., 31, 384 (1997). P. S. C. Rao, L. S. Lee, P. Nkedi-Kizza, and S. H. Yalkowsky, in Toxic Organic Chemicals in Porous Media, Z. Gerstl, Eds. Springer-Verlag, New York, 1989, Chapter 8. R. Pinal, L. S. Lee, and P. S. C. Rao, Chemosphere, 22, 939 (1991). S. H. Yalkowsky, and J. T. Rubino, J. Pharm. Sci., 74, 416 (1985). J. T. Rubino, and S. H. Yalkowsky, J. Parent. Sci. Technol., 41, 172 (1984). W. J. Lyman, W. f. Reehl, and D. H. Rosenblatt, Handbook of Chemical Property Estimation Methods: Environmental Behavior of Organic Compounds. ACS Publications, Washington, DC, 1990. C. Hansch, and A. J. Leo, Substituent constants for Correlation Analysis in Chemistry and Biology. John Wiley, New York, 1979. A. Li, and S. H. Yalkowsky, Ind. Eng. Chem. Res., 37, 4470 (1998). J. T. Rubino, and S. H. Yalkowsky, Pharm. Res., 4, 220 (1987). A. Li, and S. H. Yalkowsky, Ind. Eng. Chem. Res., 37, 4476 (1998). R. Abramowitz, Ph.D. Dissertation, University of Arizona, 1986. A. Li, and S. H. Yalkowsky, unpublished data. J. T. Rubino, Ph.D. Dissertation, University of Arizona, 1984. S. H. Yalkowsky, unpublished data. J. T. Rubino, and S. H. Yalkowsky, Pharm. Res., 4, 231 (1987). A. Fredenslund, R. L. Jones, and J. M. Prausnitz, A.I.Ch.E. J., 21, 1086 (1975). J. Gmehling, P. Rasmussen, and A. Fredenslund, Ind. Eng. Chem. Process Des. Dev., 21, 118 (1982). E. A. Macedo, U. Weidlich, J. Gmehling, and P. Rasmussen, Ind. Eng. Chem. Process Des. Dev., 22, 678 (1983). J. T. Rubino, and E. K. Obeng, J. Pharm. Sci., 80, 479 (1991). L. S. Lee, C. A. Bellin, R. Pinal, and P. S. C. Rao, Environ. Sci. Technol., 27, 165 (1997).

14.22 POLYMERS AND MAN-MADE FIBERS George Wypych ChemTec Laboratories, Inc., Toronto, Canada

The resin production industry has over 450 plants in the USA with total sales of $33 billion/year and about 60,000 employees. The man-made fiber industry has over 90 plants. It employs about 45,000 people and it has sales of $13 billion/year. In the polymer manufacture industry, production processes are diverse both in technology and equipment design. They have common steps which include preparation of reactants, polymerization, polymer recovery, polymer extrusion (if in pelletized form), and supporting operations. In some preparation operations, solvents are used to dissolve or dilute monomer and reactants. Solvent are also used to facilitate the transportation of the reaction mixture throughout the plant, to improve heat dissipation during the reaction, and to promote uniform mixing. Solvent selection is optimized to increase monomer ratio and to reduce polymerization costs and emissions. The final polymer may or may not be soluble in the solvent. These combinations of polymers and solvents are commonly used: HDPE - isobutane and hexane, LDPE - hydrocarbons, LLDPE - octene, butene, or hexene, polypropylene - hexane, heptane or liquid propylene, polystyrene - styrene or ethylbenzene, acrylic dimethylacetamide or aqueous inorganic salt solutions. These examples show that there are options available. Excess monomer may replace solvent or water can be used as the solvent. During polymer recovery unreacted monomer and solvents are separated from polymer (monomers and solvents are flashed off by lowering the pressure and sometimes degassing under vacuum), liquids and solids are separated (the polymer may be washed to remove sol-

14.22 Polymers and man-made fibers

1017

Figure 14.22.1. Schematic diagram of emissions from the polymer manufacturing industry. [Reproduced from EPA Office of Compliance Sector Notebook Project. Profile of the Petroleum Refining Industry. US Environmental Protection Agency, 1995.]

vent), and residual water and solvent are purged during polymer drying. Residual solvents are removed by further drying and extrusion. Solvents are also used in equipment cleaning. Solvents are often stored under a nitrogen blanket to minimize oxidation and contamination. When these systems are vented solvent losses occur. Figure 14.22.1 shows a schematic diagram of potential emissions during polymer manufacture. Manufacture of man-made fibers involves polymerization (usually the core part of the process), preparation of the solution, spinning, washing and coagulation, drying and other operations. Fibers are formed by forcing the viscous liquid through small-bore orifices. A suitable viscosity can be achieved either by heating or dissolution. The rheological properties of the solution are governed to a large degree by the solvents selected. Wastes generated during the spinning operation include evaporated solvent and wastewater contaminated by solvent. The typical solvents used in the production of fibers are dimethylacetamide (acrylic), acetone or chlorinated hydrocarbon (cellulose acetate), and carbon disulfide (rayon). In the dry spinning process a solution of polymer is first prepared. The solution is then heated above the boiling temperature of the solvent and the solution is extruded through spinneret. The solvent evaporates into the gas stream. With wet spinning the fiber is directly extruded into a coagulation bath where solvent diffuses into the bath liquid and the coagulant diffuses into the fiber. The fiber is washed free of solvent by passing it through an additional bath. Each process step generates emissions or wastewater. Solvents used in production are normally recovered by distillation. Figure 14.22.2 is a schematic diagram of fiber production showing that almost all stages of production generate emissions.

1018

George Wypych

Figure 14.22.2. Schematic diagram of emissions from the man-made fiber manufacturing industry. [Reproduced from EPA Office of Compliance Sector Notebook Project. Profile of the Petroleum Refining Industry. US Environmental Protection Agency, 1995.]

Tables 14.22.1 and 14.22.2 provide data on releases and transfers from both polymer manufacturing and man-made fiber production in the USA. Carbon disulfide, methanol, xylene, and ethylene glycol are used in the largest quantities. Carbon disulfide is used in manufacture of regenerated cellulose and rayon. Ethylene glycol is used in the manufacture of polyethylene terephthalate, the manufacture of alkyd resins, and as cosolvent for cellulose ethers and esters. Methanol is used in several processes, the largest being in the production of polyester. This industry is the 10th largest contributor of VOC and 7th largest in releases and transfers. There have been many initiatives to reduce emissions and usage of solvents. Man-made fiber manufacturing no longer uses benzene. DuPont eliminated o-xylene and reduced methanol and ethylene glycol use in its Wilmington operation. This change resulted in annual savings of $1 million. Process modification in a polymer processing plant resulted in a decrease in total emissions of 74% and a reduction in the release of cyclohexane by 96%. Monitoring of thousands of valves in Eastman Texas plant resulted in a program of valve replacement which eliminated 99% of the emissions. Plant in Florida eliminated solvents from cleaning and degreasing. These examples show that in many cases pollution can be reduced by better equipment, organization, and care.

14.22 Polymers and man-made fibers

1019

Table 14.22.1 Reported solvent releases from the polymer and man-made fiber industry in 1995 [Data from Ref. 1] Solvent

Amount, kg/year

Solvent

Amount, kg/year

allyl alcohol

29,000

1,4-dioxane

10,000

benzene

60,000

ethylbenzene

130,000

n-butyl alcohol

480,000

ethylene glycol

1,400,000

sec-butyl alcohol

25,000

hexane

880,000

tert-butyl alcohol

16,000

methanol

3,600,000

carbon disulfide

27,500,000

methyl ethyl ketone

260,000


100


98,000

chlorobenzene

19,000

pyridine

67,000

chloroform

14,000

tetrachloroethylene

4,000

cresol

4,000


120,000

cyclohexane

98,000

trichloroethylene

39,000

1,2-dichloroethane

98,000


12,000

dichloromethane

1,300,000

toluene

900,000


19,000

xylene

460,000

Table 14.22.2. Reported solvent transfers from the polymer and man-made fiber industry in 1995 [Data from Ref. 1] Solvent

Amount, kg/year

Solvent

Amount, kg/year

allyl alcohol

120,000

ethylbenzene

880,000

benzene

160,000

ethylene glycol

49,000,000

n-butyl alcohol

330,000

hexane

8,000,000

sec-butyl alcohol

12,000

methanol

5,600,000

tert-butyl alcohol

160,000

methyl ethyl ketone

460,000

carbon disulfide

14,000


43,000


200,000


780,000

chlorobenzene

570,000

pyridine

70,000

chloroform

59,000

tetrachloroethylene

330,000

cresol

20,000


21,000

cyclohexane

420,000

trichloroethylene

76,000

dichloromethane

250,000


98,000


300,000

toluene

2,800,000

1,4-dioxane

11,000

xylene

7,800,000

1020

George Wypych

New technology is emerging to reduce solvent use. Recent inventions disclose that, in addition to reducing solvents, the stability of ethylene polymers can be improved with the new developed process.3 A proper selection of solvent improved a stripping operation and contributed to the better quality of cyclic esters used as monomers.4 Solvent was used for the recovery of fine particles of polymer which were contaminating water.5 A new process for producing fiber for cigarette filters uses reduced amounts of solvent.6 Optical fibers are manufactured by radiation curing which eliminates solvents.7 A new electrospinning process has been developed which produces unique fibers by the dry spinning method, providing a simpler separation and regeneration of the solvent.8

REFERENCES 1 2 3 4 5 6 7 8

EPA Office of Compliance Sector Notebook Project. Profile of the Petroleum Refining Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998. M M Hughes, M E Rowland. C A Strait, US Patent 5,756,659, The Dow Chemical Company, 1998. D W Verser, A Cheung, T J Eggeman, W A Evanko, K H Schilling, M Meiser, A E Allen, M E Hillman, G E Cremeans, E S Lipinsky, US Patent 5,750,732, Chronopol, Inc., 1998. H Dallmeyer, US Patent 5,407,974, Polysar Rubber Corporation, 1995. J N Cannon, US Patent 5,512,230, Eastman Chemical Company, 1996. P J Shustack, US Patent 5,527,835, Borden, Inc., 1996. A E Zachariades, R S Porter, J Doshi, G Srinivasan, D H Reneker, Polym. News, 20, No.7, 206-7 (1995).

14.23 PRINTING INDUSTRY George Wypych ChemTec Laboratories, Inc., Toronto, Canada

The number of printing and publishing operations in the US is estimated at over 100,000. 1.5 million people are employed. The value of shipments is over $135 billion. 97% of printing is done by lithography, gravure, flexography, letterpress, and screen printing on substrates such as paper, plastic metal, and ceramic. Although, these processes differ, the common feature is the use of cleaning solvents in imaging, platemaking, printing, and finishing operation. Most inks contain solvents and many of the adhesives used in finishing operations also contain solvents. Many processes use the so-called fountain solutions which are applied to enable the non-image area of the printing plate to repel ink. These solutions contain primarily isopropyl alcohol. But the printing operation is, by itself, the largest contributor of VOCs. Each printing process requires inks which differ drastically in rheology. For example, gravure printing requires low viscosity inks which contain a higher solvent concentrations. Tables 14.23.1 and 14.23.2 provide data on the reported releases and transfers of solvents by the US printing industry. These data show that there are fewer solvents and relatively low releases and transfers compared with other industries. In terms of VOC contribution, the printing industry is 5th and 10th in the total emissions and transfers. Current literature shows that there is extensive activity within and outside industry to limit VOCs and reduce emissions. Cleaning operations are the major influence on emissions. Shell has developed a new cleaning formulations containing no aromatic or chlori-

14.23 Printing industry

1021

nated hydrocarbons.3 An additional requirement was to optimize the solvent mixture to prevent swelling of the rubber in blanket cylinders and rollers. It is predicted that the European industry will increase rate of the introduction of radiation-cured inks and eliminate isopropanol from fountain solutions.4 It is expected that radiation-cured flexographic inks will grow by 30%/year in the next five years.5 In Germany, 70-80% of emissions or 47,000 ton/year will have to be eliminated by the year 2007.6 Beginning in 1999, the UK industry must keep VOC concentration below 5 tonnes/year per plant.7 VOC concentration in outside atmosphere must not exceed 150 mg/m3 (50 mg/m3 if there is more than 5% aromatic solvents). Reactive hot melts are being used in book binding.8 This will eliminate emissions from currently used solvent adhesives. Solvent replacements are not the only solution at hand. Solvent-containing systems often give better quality than replacement systems, therefore methods have been developed to make the solvent based materials more acceptable. A soil bed biofiltration system was tested in California with excellent results.9 This biofilter is a bed of soil impregnated with microorganisms which use VOC as their food. Present California regulations require that such a treatment system has a 67% capture and VOC destruction efficiency. The new method was proven to have 95.8% efficiency. In addition to environmental issues with solvents, the printing industry has addressed the source of their raw materials. Present systems are based on petroleum products which are not considered renewable resources. Terpenes are natural products which are now finding applications in the print industry.10 In Denmark, of 70% cleaning solvents are vegetable oil based. These and other such innovations will continue to be applied to reduce solvent use and emissions. It is also reported10 that water-based system replaces fountain solutions. Other factors are driving changes. Odors in packaging materials and the migration of solvent to foods are unacceptable. Most odors in packaging materials are associated with process and coalescing solvents.11 Foods which do not contain fat are more susceptible to the retaining the taste of solvents. Printing inks which may be acceptable for foods containing fat may not be suitable for fat-free applications (see more on this subject in Chapter 16.1).12 Many recent inventions have also been directed at solving the current environmental problems of printing industry.13-21 The solvent in gravure printing inks not only contribute to pollution but also to the cost of solvent recovery and/or degradation. A new technology is proposed in which a solvent free ink with a low melting point can be processed in liquid state and then be solidified on cooling.13 A non-volatile solvent for printing inks was developed based on a cyclic keto-enol tautomer and a drying oil.14 An alcohol soluble polyamide for rotary letterpress printing inks was developed15 and subsequently adapted to flexographic/gravure inks.18 A polyamide was also used in a rotary letterpress ink which enabled low alcohols to be used as the solvent with some addition of an ester.16 This new ink is compatible with water-based primers and adhesives which could not be used with solvent-based inks. Inks for jet printers are water sensitive. One solvent-based technology was developed using esters and glycols17 and the other using low alcohols.20 Another recent invention describes aqueous ink containing some low alcohols.19 UV and electron beam cured ink concentrates were also developed.21 This information from open and patent literature clearly indicates that industry is actively working on the development of new technological processes to reduce emissions of solvents.

1022

George Wypych

Table 14.23.1. Reported solvent releases from the printing and publishing industry in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

n-butyl alcohol

43,000


170,000

dichloromethane

59,000


31,000

1,4-dioxane

8,000

tetrachloroethylene

34,000

ethylene glycol

46,000


180,000

ethylbenzene

23,000

trichloroethylene

13,000

hexane

50,000


36,000

isopropyl alcohol

27,000

toluene

12,200,000

methanol

170,000

xylene

700,000

methyl ethyl ketone

960,000

Table 14.23.2. Reported solvent transfers from the printing and publishing industry in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

n-butyl alcohol

7,000


63,000

dichloromethane

43,000


28,000

1,4-dioxane

340

tetrachloroethylene

27,000

ethylene glycol

16,000


39,000

ethylbenzene

9,200

trichloroethylene

4,000

hexane

12,000


33,000

isopropyl alcohol

12,000

toluene

2,800,000

methanol

17,000

xylene

240,000

methyl ethyl ketone

700,000


EPA Office of Compliance Sector Notebook Project. Profile of the Printing and Publishing Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998. N C M Beers, M J C M Koppes, L A M Rupert, Pigment & Resin Technol., 27, No.5, 289-97 (1998). D Blanchard, Surface Coatings International, 80, No.10, 476-8 (1997). B Gain, Chem. Week, 160, No.14, 28-30 (1998). W Fleck, Coating, 31(1), 23-25 (1998). C H Williams, Converter, 34, No.9, 11-2 (1997).

14.24 Pulp and paper

8 9 10 11 12 13 14 15 16 17 18 19 20 21

1023

Hughes F, TAPPI 1997 Hot Melt Symposium. Conference Proceedings. TAPPI. Hilton Head, SC, 15th-18th June 1997, p.15-21. A Mykytiuk, Paper, Film & Foil Converter, 72, No.8, 120-3 (1998). A Harris, Paper, Film & Foil Converter, 72, No.5, 198-9 (1998). R M Podhajny, Paper, Film & Foil Converter, 72, No.12, 24 (1998). T Clark, Paper, Film & Foil Converter, 70, No.11, 48-50 (1996). R Griebel, K A Kocherscheid, K Stammen, US Patent 5,496,879, Siegwerk Druckfarben GmbH, 1996. D Westerhoff, US Patent 5,506,294, 1996. P D Whyzmuzis, K Breindel, R A Lovald, US Patent 5,523,335, Henkel Corporation, 1996. R J Catena, M C Mathew, S E Barreto, N Marinelli, US Patent 5,658,968, Sun Chemical Corporation, 1997. J M Kruse, US Patent 5,663,217, XAAR Ltd., 1997 P D Whymusis, US Patent 5,714,526, Henkel Corporation, 1998. H Yanagi, S Wakabayashi, K Kaida, US Patent 5,736,606, Kao Corporation, 1998. M Shinozuka, Y Miyazawa, M Fujino, T Ito, O Ishibashi, US Patent 5,750,592, Seiko Epson Corporation, 1998. W R Likavec, C R Bradley, US Patent 5,866,628, Day-Glo Color Corporation, 1999.

14.24 PULP AND PAPER George Wypych ChemTec Laboratories, Inc., Toronto, Canada

The US pulp and paper industry operates over 550 facilities which employ over 200,000 people. Total shipments are $60 billion with an additional $80 billion in converted products. Several processes contribute to the emission of solvents. These include chemical pulping kraft process (terpenes, alcohols, methanol, acetone, chloroform), bleaching (acetone, dichloromethane, chloroform, methyl ethyl ketone, carbon disulfide, chloromethane, and trichloroethane), wastewater treatment (terpenes, alcohols, methanol, acetone, chloroform and methyl ethyl ketone), and evaporators in chemical recovery systems (alcohols and terpenes).

Table 14.24.1. Reported solvent releases from the pulp and paper industry in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

320,000

methanol

63,000,000

n-butyl alcohol

46,000

methyl ethyl ketone

700,000

chloroform

4,500,000


10,000

chloromethane

260,000


17,000

cresol

410,000

toluene

580,000

ethylbenzene

22,000

xylene

49,000

ethylene glycol

37,000

o-xylene

260

hexane

150,000

1024

George Wypych

Table 14.24.2. Reported solvent transfers from the pulp and paper industry in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

24,000

hexane

8,600

n-butyl alcohol

16,000

methanol

23,000,000

chloroform

150,000

methyl ethyl ketone

36,000

chloromethane

120


1,400

cresol

3,600

toluene

23,000

ethylene glycol

190,000

xylene

4,000

Table 14.25.1. Reported solvent releases from the rubber and plastics industry in 1995 [Data from Ref. 2] Solvent benzene n-butyl alcohol sec-butyl alcohol tert-butyl alcohol carbon disulfide chlorobenzene chloroform chloromethane cresol cyclohexane dichloromethane N,N-dimethylformamide 1,4-dioxane ethylbenzene

Amount, kg/year 5,800 380,000 17,000 240 5,500,000 5,000 46,000 47,000 9,000 480,000 11,700,000 350,000 2,600 210,000

Solvent ethylene glycol hexane isopropyl alcohol methanol methyl ethyl ketone methyl isobutyl ketone N-methyl-2-pyrrolidone tetrachloroethylene 1,1,1-trichloroethane trichloroethylene 1,2,4-trimethylbenzene toluene xylene m-xylene

Amount, kg/year 120,000 1,700,000 28,000 4,000,000 5,500,000 1,100,000 32,000 160,000 3,000,000 660,000 260,000 7,600,000 2,200,000 6,000

Tables 14.24.1 and 14.24.2 give the reported releases and transfers of solvent data for the US pulp and paper industry. If not for the emissions of methanol and chloroform the industry would be a much less serious polluter. It is 7th in VOC contributions and 8th in total releases and transfers.

REFERENCES 1 2

EPA Office of Compliance Sector Notebook Project. Profile of the Pulp and Paper Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.

14.25 Rubber and plastics

1025

Table 14.25.2. Reported solvent transfers from the rubber and plastics industry in 1995 [Data from Ref. 2] Solvent

Amount, kg/year

Solvent

Amount, kg/year

benzene

15,000

hexane

50,000

n-butyl alcohol

370,000

isopropyl alcohol

14,000

sec-butyl alcohol

1,100

methanol

1,400,000

tert-butyl alcohol

85,000

methyl ethyl ketone

3,500,000

carbon disulfide

150,000


450,000

chloroform

1,200


120,000

chloromethane

330

tetrachloroethylene

47,000

cresol

3,000


160,000

cyclohexane

350,000

trichloroethylene

170,000

dichloromethane

900,000


14,000


570,000

toluene

2,200,000

1,4-dioxane

49,000

xylene

940,000

ethylbenzene

350,000

m-xylene

5,700

ethylene glycol

15,300,000

14.25 RUBBER AND PLASTICS George Wypych ChemTec Laboratories, Inc., Toronto, Canada

The US rubber and plastics industry employs over 800,000 people and operates over 12,000 plants. Its total production output is estimated at over $90 billion. The industry produces a wide diversity of products some of which do not contain solvents but many of which require the use of process solvents. Solvents are contained in adhesives used in finishing operations. Large quantities of solvents are used for surface cleaning and cleaning of equipment. Tables 14.25.1 and 14.25.2 provide data on the reported releases and transfers of solvents by the US rubber and plastics industry. These industries contribute small amounts of VOC which are in the range of 0.00001-0.00005 kg VOC/kg of processed rubber. It was the ninth largest contributor to releases and transfers of all US industries. Dichloromethane, toluene, carbon disulfide, methyl ethyl ketone, methanol, 1,1,1-trichloroethane, hexane, methyl isobutyl ketone, and xylene are emitted in very large quantities.

REFERENCES 1 2

EPA Office of Compliance Sector Notebook Project. Profile of the Rubber and Plastics Industry. US Environmental Protection Agency, 1995. EPA Office of Compliance Sector Notebook Project. Sector Notebook Data Refresh - 1997. US Environmental Protection Agency, 1998.

1026

Mohamed Serageldin, Dave Reeves

14.26 USE OF SOLVENTS IN THE SHIPBUILDING AND SHIP REPAIR INDUSTRY Mohamed Serageldin U.S. Environmental Protection Agency, Research Triangle Park, NC, USA

Dave Reeves Midwest Research Institute, Cary, NC, USA

14.26.1 INTRODUCTION The focus of this chapter will be on the use of solvents in the shipbuilding and ship repair industry. This industrial sector is involved in building, repairing, repainting, converting, or alteration of marine and fresh water vessels. These vessels include self-propelled vessels, those propelled by other vessels (barges), military and Coast Guard vessels, commercial cargo and passenger vessels, patrol and pilot boats, and dredges. The industry sector is also involved in repairing and coating navigational aids such as buoys. This chapter begins with an overview of operations in a typical shipbuilding and/or ship repair facility (shipyard), to identify those operations that generate significant volatile organic compound (VOC) emissions and/or hazardous air pollutant (HAP) emissions from the use of organic solvents. Organic solvents that are VOCs contribute to formation of ozone in the troposphere. Other organic solvents such as chlorinated fluorocarbons (CFCs) cause depletion of the ozone layer in the stratosphere. Therefore, VOCs and other air toxics, such as those compounds listed as HAPs, are both indirectly and directly detrimental to the general public’s health. Because many solvents are VOCs and often contain large amounts of HAPs, many state agencies1,2 and the United States Environmental Protection Agency (U.S. EPA) have issued regulations to limit their content in materials used for surface coating and cleaning operations at shipyards.3-7 14.26.2 SHIPBUILDING AND SHIP REPAIR OPERATIONS Most facilities engaged in shipbuilding or ship repair activities (shipyards) have several manufacturing areas in common, each including one or more “unit operations”. These areas include: (a) surface preparation of primarily steel surfaces, which may include cleaning with multiple organic solvents; (b) assembly operations, which involve assembly of blocks that were constructed from sub-assembled parts (this step involves steel cutting and material movement using heavy equipment such as cranes); (c) cleaning operations (other than surface preparation) such as equipment and parts cleaning; and (d) coating operations.8,9 There are secondary operations such as chrome plating, asbestos removal, fuel combustion, carpentry, and, to various degrees, polyester lay-up operations (composite materials construction activities). We will next discuss those operations that involve the use of organic cleaning solvents. 14.26.3 COATING OPERATIONS Marine coatings can be applied by the use of spraying equipment, brushes, or rollers. Coating operations at shipyards are typically conducted at two primary locations: (1) outdoor work areas or (2) indoor spray booths. The outdoor work areas can include ship exteriors and interiors. Most shipyards report that typically only a small percentage (10%) of the coating operations are done indoors. However, in large construction yards a larger propor-

14.26 Use of solvents in the shipbuilding

1027

tion (up to 30 %) of the coatings are applied indoors.10 Coating and cleaning operations constitute the major source of VOC and HAP emissions from shipyards. If the metal surface is not well prepared before a coating is applied or if the coating is applied at the wrong ambient conditions, the coating system may fail and the work may have to be redone. The amount of cleaning necessary will depend on the type and extent of the problem and the coating system that is being used. 14.26.4 CLEANING OPERATIONS USING ORGANIC SOLVENTS In most industrial applications involving metal substrates, organic cleaning solvents are used to remove contaminants or undesirable materials from surfaces before a coating is applied to clean equipment and parts utilized to apply the coating or soiled during that operation. Solvents are used for general maintenance of equipment parts. These surfaces are typically made of steel. However, vessels may also be made from natural materials such as wood and synthetic materials such as fiberglass. Therefore, a solvent must be selected that will not attack the substrate being cleaned. For material accounting purposes, we can classify cleaning (unit) operations as follows:11,12 1. Surface preparation of large manufactured components (stage before a coating is applied). 2. Surface preparation of small manufactured components (stage before a coating is applied). 3. Line cleaning (includes piping network and any associated tanks). 4. Gun cleaning (manually or in a machine). 5. Spray booth cleaning (walls and floor). 6. Tank cleaning (mostly inner tank surfaces and any associated pipes). 7. Parts (machine) cleaning (simple dip tanks and large machines). 8. Cleaning of equipment and other items (e.g., bearings, buckets, brushes, contact switches). 9. Floor cleaning (organic solvents are no longer used). These categories are similar to those found in other industries involved in the application of surface coating. However, the number of cleaning categories varies from one industry to another. For example, the automotive manufacturing industry (SIC code 3711) and the furniture industry are involved to various degrees in all nine types of cleaning operations. On the other hand, the photographic supplies (chemicals) industry will not include the first three listed cleaning operations.11 14.26.4.1 Surface preparation and initial corrosion protection Large manufactured ship components are often cleaned with an organic solvent as the first of a number of cleaning steps that are required before a coating is applied. The method of surface preparation is selected to work with a chosen coating system. Surface preparation may include application of chemicals such as etching agents, organic solvents cleaners, and alkaline cleaners. Organic solvents such as mineral spirits, chlorinated solvents, and coal tar solvents are used to remove unwanted materials such as oil and grease.13 If a ship is being repaired, existing coatings usually need to be removed. Solvents such as dichloromethane are commonly used for removing (stripping off) old or damaged coatings. However, aqueous systems involving caustic compounds are now being used more frequently for such purposes.14 Pressure washing and hydro blasting are other cleaning techniques used. But, the

1028


predominant method is still particulate blasting (using abrasive media), which is used to remove mil scale, extra weld material, rust, and old coatings. The angle at which the surface is blasted is chosen to generate the desired peaks and valleys on the substrate, that will accommodate the viscosity, chemistry (polar groups) of the primer coating. The surface profiling will also help the primer coating adhere mechanically to the substrate, contributing to the longevity of the coating system.15 Pre-construction primers are sometimes used immediately following surface preparation (blasting) to prevent steel from oxidizing (rusting). This primer is removed by particulate blasting, before the protective coating system (one or more coatings) is applied to the assembled parts or blocks. Removal of such primers (when they cannot be welded-through) can result in emissions of VOCs and HAPs. 14.26.4.2 Cleaning operations after coatings are applied Surface coating operations at shipyards use predominantly solvent-based coatings. Hence, relatively large amounts of organic solvents are used for cleaning and thinning activities. Table 14.26.1 shows the most common organic solvents used for thinning and cleaning, based on 1992 data.16 Table 14.26.2 gives examples of solvent products that can be used for both thinning coatings and for cleaning surfaces after coatings are applied and for maintenance cleaning. The solvent products are listed in decreasing order of evaporative rate. Acetone, a ketone solvent is commonly used for cleaning and thinning polyester resins and gel coats. However, it is also used in formulating low-VOC and low-HAP products. Methyl ethyl ketone (MEK) and methyl isobutyl ketone (MIBK) are fast evaporative solvents that are used for thinning and cleaning vinyl coatings, epoxy coatings, and many other high performance coatings. Fast evaporative coatings that can improve application properties for a good finish may also be formulated by blending different solvents. Examples are shown in Table 14.26.2. The fast evaporative mix includes solvents varying in polarity and solubility parameters. They include an oxygenated solvent (MIBK), aromatic hydrocarbon solvents that contain less than 10 percent (by mass) HAPs, and aromatic hydrocarbons like xylene that are 100 percent HAPs as will be shown later. Together they produce the correct solvency for the polymer (resin). Table 14.26.1. Predominant solvents used in marine coatings [from ref. 16 ] and EPA regulatory classifications Organic solvent

VOC

HAP, Sec. 112 (d)

Toxic chemicals, Sec. 313

Butyl alcohol

Y

Y

Y

Ethyl alcohol

Y

N

N

Isopropyl alcohol

Y

N

Ya

Xylene

Y

Y

Y

Toluene

Y

Y

Y

Ethyl benzene

Y

Y

Y

ALCOHOLS

AROMATICS


Organic solvent

1029

VOC

HAP, Sec. 112 (d)

Toxic chemicals, Sec. 313

Ethylene glycol ethers

Y

Y

Y

Propylene glycol ethers

Y

N

N

Acetone

N

N

Y

Methyl ethyl ketone

Y

Y

Y


Y

Y

Y

Methyl amyl ketone

Y

N

N

Y

Yb

N

High-flash naphtha

Y

Y

b

Y

n-Hexane

Y

Y

N

ETHERS

KETONES

PARAFFINIC Mineral spirits

VOC = volatile organic compound; HAP = Hazardous air pollutant; Sec 313 of the Emergency Right-to-know Act (EPCRA), also known as Title III of the Superfund Amendments and Reauthorization Act of 1986 (40 CFR Part 372). aUse of strong acid process, no supplier notification. bLigroine (light naphtha), VM&P naphtha, Stoddard solvent, and certain paint thinners are also commonly referred to as mineral spirits. These distillation fractions contain less than 10 % by mass HAPs (see Table 14.26.4).

Vapor press., mmHg @ 20oC

Surface tension, dynes/cm 20oC

Av. sp. gr, @ 25oC

Viscosity, cP @ 25oC

6.1

186

56

27.1

0.787

0.31

MEK*

Vinyl

9.3

4.0

70

80

24.2

0.806

0.43

MIBK**, 24 %

8.58

1.7

28

116

23.3

0.796

0.54

N-butyl alc., 24%

11.6

0.44

5.5

118

23.4

0.806

2.62

Toluene, 52%

8.93

2.0

22

111

28.2

0.863

0.57

MIBK, 23 %

8.58

1.7

28

116

23.3

0.796

0.54

EGBE***, 26%

10.2

0.072

0.6

>169

26.9

0.899

3.0

AHC****, 30%

7.7

0.16

2.0

>160

23.4

0.775

0.88

1,2,4-Trimethylbenzene, 16%

8.9

19

2.1

168

30.2

0.871

0.94

Xylene (mixed), 2%

9.9

0.77

6.0

>135

27.6

0.856

0.63

Spraying thinner & solvent Epoxy b cleaner (fast)

Brushing thinner & solvent cleaner b (medium)

Epoxy

MEK, 100 % (approx.)

B.P., oC @ 760 mmHg

Relative rate, nBUOAc =1.0

9.8

Compound, wt%

Polyester Acetone, 100% (approx.)

Typical coating

Acetone

Thinner & cleaning solvents

Solubility parameter, (cal/cm3)1/2

Table 14.26.2. Selected products that are used as both solvent thinners and solvent cleaners

B.P., oC @ 760 mmHg

Surface tension, dynes/cm 20oC

Av. sp. gr, @ 25oC

0.072

0.6

>169

26.9

0.899

3.0

AHC, 30%

7.7

0.16

2.0

>160

23.4

0.775

0.88

1,2,4-Trimethylbenzene, 16%

8.9

19

2.1

168

30.2

0.871

0.94

Xylene (mixed), 1%

9.9

0.77

6.0

>135

27.6

0.856

0.63

Relative rate, nBUOAc =1.0

10.2

Solubility parameter, (cal/cm3)1/2

EGBE, 51%

Compound, wt%

Vapor press., mmHg @ 20oC

Lacquer retarder (thinLacquer ner) & cleaner b (slow)

Viscosity, cP @ 25oC


Typical coating

Thinner & cleaning solvents

1030

*MEK=methyl ethyl ketone, **MIBK=methyl isobutyl ketone, ***EGBE=ethylene glycol monobutyl ether, ****AHC=aromatic hydrocarbon solvent; aPhysical properties mainly from Industrial Solvents Handbook, 110 -114. bMaterial Safety Data Sheet (Mobile Paint Co. Alabama)

The lower specific gravity of ketones (see Table 14.26.2) than other materials such as glycol ethers helps reduce total mass of VOCs (or HAPs) per volume of nonvolatiles (solids) in a container of coating. Glycol ethers are good solvents for epoxies and acrylics. They also have good coupling abilities in blends of poorly miscible solvents17 and have low evaporative rates. The properties of a solvent product are dependent on the chemical structure and distillation range of the solvent mix in the product. The latter will affect the evaporative rate from a coating or cleaner, affecting the solubility of the resin in the coating and viscosity of the coating and solvent cleaner. Therefore, the viscosity of the solvent product must be close to that of the resin in a coating.18 The surface tension of a solvent provides a measure of the penetrability of a cleaning solvent. A low surface tension also means the solvent spreads more readily, which is an important property for a cleaning product. However, several properties in Table 14.26.2 come into play in determining the effectiveness of a cleaning solvent. Most coating operations, due to the size and accessibility of ships, occur in open air in drydocks, graving docks, railway, or other locations throughout a facility. Because of the size of ships, the predominant application method is airless spray guns. The thickness of the coating will determine if the application equipment needs to be cleaned during application of the coating or after the job is completed. The lines from the supply tanks to the spray gun may in some instances exceed 46 m (150 ft) in length. The ensemble of equipment and items that have to do with the application of the coating or “unit operation system (UOS)” is shown schematically in Figure 14.26.1. The representation depicts a layout for outdoor application of coatings. It includes the container used to hold the coating, attached feed pump, line transferring the coating to the spray gun, the spray gun itself, and any other item soiled with a coating that will need to be cleaned with organic solvent before it can be reused. The need and frequency for cleaning will depend on the individual facility or company cleanliness standard (i.e., requirements) and the number of coating formulation or color changes. Cleaning of spray guns, internal transfer lines, and associated tanks account for a large part of organic solvent usage. At most shipyards, a small percentage of the coatings are applied indoors, in spray-booths. The walls and floors of these booths are cleaned by wiping with a solvent laden cloth. The coating application equipment UOS for most facilities will look very similar to that shown in Figure 14.26.1, except that the coating transfer


1031

lines will be shorter if the coating storage tanks are positioned close to the spray booths. The transfer lines, that will need to be cleaned with solvent, will be longer if the coating tanks are located away from the application area. When this is the case, the transfer lines typically run underground at the facility and another representation than the one shown in Figure 14.26.1 will need to be used, to clearly identify the emission points and waste streams for properly quantifying solvent losses. The latter may include a unit for recycling or reclaiming solvents. Spray gun cleaning procedures may be a once-through type with collection of spent solvent in a container for disposal or reuse. Some facilities use commercial gun washers. Because gun washers are enclosed and recirculate solvent, they can reduce the amount of solvent lost by evaporation. In either case, the emissions are calculated as the difference between the amount used and the amount recovered. To calculate the emissions associated with cleaning a spray gun it is recommended that a material balance around a “unit operation system” be considered. Several examples are provided in the Alternative Control Techniques (ACT) document on industrial cleaning solvents.19 Several types of part cleaners are used at shipbuilding and ship repair facilities. The types used in such facilities vary from the more simple sink and spray systems20 to more elaborate parts (machine) cleaners of the cold or vapor types.4 Most of the parts cleaners in shipyards are small — around 1.5 m x 1 m and 1 m deep - usually located in the machine shops, not the paint rooms. Most of the parts are small components being cleaned prior to being joined to other small parts into assemblies and sub-assemblies or being cleaned as part of some type of repair operation. Most of parts cleaners used were basket-type design with the parts loaded into a basket and dropped through the vapor zone several times to clean off the oils and dirt. Some shipyards use contractors to come in and change out the solvent on a routine schedule. 14.26.4.3 Maintenance cleaning of equipment items and components Shipyards also undertake scheduled maintenance cleaning of many ship components such as contacts and switches and equipment items such as bearings and packaging machines. This is mostly done by hand-wiping the parts with organic solvents. These operations will generally consume a relatively small amount of the overall volume of organic solvents used for cleaning in shipyards. Solvents are also used in machine shop areas and thus contribute to the waste stream. 14.26.5 MARINE COATINGS There are several categories of marine coatings that are used to protect the surface of a ship from the aggressive marine environment Figure 14.26.1 Schematic diagram of marine coating application equipment.

1032


and for other performance requirements such as preventing corrosion and fouling; protecting cargoes from contamination; providing safety warnings and informational markings; providing cosmetic and camouflage colors; preventing slipping and sliding on walking surfaces; reducing fire hazards; and providing cathodic protection.21 The coating systems of marine coatings are selected to meet: • the type of marine environment to which a vessel will be exposed • the time a vessel is to remain operational before it needs to be reworked. General areas of a ship include: (1) underwater hull, (2)superstructures and freeboard, (3) interior habitability areas, (4) exterior deck areas, and (5) fuel, water ballast and cargo tank.22 The freeboard is the area above water hull. These areas have different characteristics and operational requirements. Table 14.26.323 shows the predominant resin and solvent types used on ships based on a 1991/1992 survey of the industry obtained as part of the shipbuilding and ship repair regulation was being developed. The summary table also gives average VOC and HAP content for the various coating category types. Epoxy coatings constitute a large percentage of the coatings used. The epoxy films are strongly resistant to most chemicals and are very good anti-corrosion coatings, and require little surface preparation. Table 14.26.3. Summary of marine coating usage (by coating type)23 Coating types

Average usage in U.S. shipyards, %

Average VOC content, g/L (lb/gal)

Average HAP content, g/L (lb/gal)

General use types alkyd based

10

474

(3.95)

355

(2.98)

epoxy based

59

350

(2.92)

56

(0.47)

antifouling (multiple resins)1

11

388

(3.23)

268

(2.25)

inorganic zinc based

10

545

(4.54)

274

(2.30)

other speciality categories

10

400

(3.33)

144

(1.20)

TOTAL

100

Speciality types

1

resins: epoxy, polyurethane, vinyl, and chlorinated rubber

The coating system used will depend on service requirements. Maximum protection at an economical price can be achieved when the user understands the protection needed and the functions performed by the coatings. Coatings are designed for spray viscosity, drying time, pot-life, and cure profile; all of these parameters affect shelf stability.24 The physical parameters and properties of a coating are affected by the volatile constituents (mainly organic solvents) in a coating, some of which are VOCs, HAPs, ozone depleters, and SARA 313 toxic chemicals that need to be reported.25 14.26.6 THINNING OF MARINE COATINGS The sprayability of a coating is determined by its viscosity at application. The viscosity is a measure of the ability of a material to resist flow and is an important formulation design parameter. Application viscosity is affected by the ambient conditions and by the degree of


1033

mixing that occurs within the applicator. The thinner is often the same material as the cleaner (as indicated in Table 14.26.2). The solvent material is often a blend of miscible materials. Together they will dissolve a dry resin that needs to be removed or give the coating certain needed properties such as reduced/increased viscosity or shorter drying times. Standard spraying equipment will apply coatings up to some maximum viscosity. Above that maximum value, thinning solvents are required. Thinning solvent is sometimes added to enhance brushability or sprayability of a coating. The appropriate viscosity is provided by the coating manufacturer or supplier; it will depend on the solvent content of the coating and temperature at the point of application.26 Since most coatings are applied outdoors, extreme weather conditions may require adding thinning solvents to the coating. Organic thinning solvents are added to coatings to alter their flowing properties. However, the flow properties of a coating may be altered by using special heaters or a combination of solvent and heat. The effect of a heater on the viscosity of a coating depends on the physical properties of the coating and on the flow rate in the in-line heater. Under cold weather conditions, in-line heaters may provide good viscosity control, but may not be able to solve all application problems that are encountered in the field. Under extremely low temperatures, the substrate surface can act as a heat sink, which may inhibit the setting or curing of the coating. In-line heaters which are used for low volume coatings are not suitable for large volume coatings. As a result, thinning solvents are still needed to transfer the fluid from storage to pumps and hoses. Under hot and humid weather conditions, certain coatings (e.g., lacquers) can rapidly lose organic solvent prior to and during application. Often under these situations a facility will add solvent blends to make up for the reduction in viscosity and to overcome condensation on the surface (blushing).27 Evaporative losses can be minimized by adopting good work practices and by using formulations that contain organic solvents with low vapor pressures. 14.26.7 SOLVENT EMISSIONS Several states with their own rules regulating marine coatings have separate rules addressing solvent cleaning operations. While marine coating rules typically address VOC contents and types of application equipment, the cleaning solvent rules are more generic and address cleaning solvents used at any and all metal-related manufacturing operations. Many types of solvents are used in marine coatings and in their associated cleaning materials as shown in Table 14.26.1. Almost all solvents used at shipyards are VOC and approximately one in three solvents contain HAPs. Of the HAPs reported, several are included on the list of 17 high priority chemicals targeted by U.S. EPA for the 33/50 program.28 These included xylene (commercial), toluene, and the ketones. Commercial grade xylene represents the major portion of the volatile HAPs reported. Many of the commonly known solvents are actually petroleum distillation fractions and are composed of a number of compounds (e.g., mineral spirits and naphthas). There are two general types of solvents derived from petroleum, aliphatics or aromatics. Aromatics are stronger solvents than aliphatics since they dissolve a wider variety of resins. Most major solvent suppliers (chemical manufacturers) produce several types and variations of these solvents and the associated HAP contents can vary significantly from manufacturer to manufacturer and from batch to batch. These types of solvents are used extensively and are present in the majority of marine coatings. Table 14.26.4 provides a summary of common petroleum distillate solvents and solvent blends and their associated HAP content. For any

1034


solvent or solvent blend that is not listed as specified in Table 14.26.4, another table (Table 14.26.5) was developed to provide solvent groupings and associated HAP component/content values. The HAP values for Tables 14.26.4 and 14.26.5 were adapted from estimates provided in 1998 by the Chemical Manufacturer Association’s Solvent’s Council. Table 14.26.4. HAP content of single solvents and solvent blends. [Adapted, by permission from Chemical Manufacturer Association’s Solvent’s Council] Solvent/solvent blend

CAS No.

HAP content range, wt%

Average HAP content, wt%

Typical HAP, wt%

Toluene

108-88-3

100

100

toluene

Xylene(s)

1330-20-7

100

100

xylenes, ethylbenzenea

Hexane

110-54-3

49-55

50

n-hexaneb

Ethylbenzene

100-41-4

100

100

ethylbenzene

Aliphatic 140

0

0

none

Aromatic 100

1200

The national names after the antibiotics indicate the country the antibiotics originated in.

15

Inhibitors of electron transport dinitrophenol

4

34.3

17

FCCP

19

0.20

36

potassium cyanide

17

12.2

48

sodium azide

17

630

19

15.2.2 A simple test to determine toxicity using bacteria

n

Ave.

1103

Var.

n

Ave.

Var.

atropine

9

191

11

Herbicides continuation

Alkaloids

Metribuzin

6

>1200

Naproamide

10

289

43

caffeine

3

3700

18

Nicosulfuron

10

267

11

immitine

11

122

8

Norflurazon

8

182

14

nicotine

13

990

13

Oxadiazon

6

269

16

quinine

4

131

14

Quniclorac

6

>1200

qunidine

4

137

15

Sethoxydim

6

2.70

17

scopolomine

5

129

5

Thiazopyr

6

43.2

4

Thifensulfuron

7

928

4

Trifluralin

7

10.3

18

n = the number of samples tests; ave. = the average of the values obtained, reported as part per million (ppm), mg per liter; var. = the variance in the results, the standard deviation divided by the mean.

15.2.2.5 Comparisons with other tests The Rhizobium test has been compared with other tests. This is done by finding values in the literature for various toxins using different tests and plotting the log of these values vs. the log of values from the second assay. A regression line was plotted and the correlation coefficient calculated from the regression coefficient (the correlation coefficient is the square root of the regression coefficient). In Figure 15.2.2.2, values for chemicals from two laboratories

Figure 15.2.2.2. Representative plot of the comparison of two assays. These are data from Daphnia 1 (Calleja et al., 1993) and Daphnia 2 (Lilius et al.,1994).

1104

James L. Botsford

(Calleja et al., 1993; Lilius et al., 1994) using the assay involving the sand flea Daphnia magna are presented. If two systems give the same results, the correlation coefficient will be 1.00. In the literature there are reports of the toxicity of 50 MEIC chemicals (MEIC, multicenter evaluation of cytotoxicity) tests used in Europe to evaluate different tests for toxicity. When these two tests with Daphnia were compared, a correlation coefficient of 0.895 was obtained (Figure 15.2.2.2). This indicates that when a standardized test is run in two laboratories with different personnel, the results are not identical. This discrepancy is noted in tests involving animals. Apparently variations in the strains of animals, differences in care and other factors influence the results of toxicity tests with animals. Often variations of 10x are observed. This sort of analysis was carried using the Rhizobium assay. Published values for the chemicals using Microtox, LD50 reports for rats, IC50 (testing with animal cells), Daphnia magna, and HLD (Human Lethal Dose determined from autopsy reports) are included (Table 15.2.2.2). These values for the assay with Rhizobium will permit comparisons to be made (Table 15.2.2.3). It should be noted that the values for rats and HLD are much higher than with the other methods. With animal tests, the animals are force fed the chemical, it is injected through a tube into their stomach. Occasionally the toxin is injected intraperitoneally. It is uncertain what happens to the toxic chemical in the acidic stomach. This methodology has been criticized because humans are rarely exposed to toxic chemicals in this fashion. HLD data comes from autopsy reports and it can never be certain how much of the chemical the victim actually came in contact with, it can only be estimated. And this method only provides data for chemicals already in the environment, it is not a good method to predict toxicity.

rat hepat, m moles

HLD, m moles

Daphnia, m moles

IC50, m moles

Rat, m moles

Microtox, m moles

Chemical tested

Rhizobium, m moles

Table 15.2.2.2. MEIC chemicals tested. Comparison

acetomenaphen

1.649

2.19

15.8

1.45

0.269

1.698

10.75

acetylsalicylic acid

1.106

0.145

1.122

1.7

0.932

2.818

2.662

amitriptyline

0.0146

0.078

1.148

0.1

0.02

0.2

0.07

barium chloride

0.109

caffeine

1.02

3.388

1

1.58

3.388

1

1.596


1.91

4.786

15.1

22.1

126

0.447

3.984

chloroform

5.29

12.9

7.586

2.63

1.738

6.198

chlororamphenicol

4.332

1.122

7.7

copper II sulfate

0.007

0.01

1.86

0.001

0.316

0.048

dichloromethane

4.122

37.1

18.6

10.5

4.17

109.1

digoxin

0.231

ethanol

1643

ethylene glycol*

3427

1.349

0.794

0.54

0.47

5.248

0.402

0.426

0.12

12.8

0.0014

0.271

691

151

158

234

109.6

451

1778

75.8

322

1202

25.7

358

0.138

0.0068

0.00027

isopropyl alcohol

950

380

83.2

90.5

155

41.7

304

iron II sulfate

0.32

0.782

2.089

0.302

3.020

1.621

lindane

0.141

21.9

0.263

malithion

0.112

0.479

0.871

0.00006

0.0002

3.715

methanol

2130

912

178

nicotine

6.1

0.224

0.309

orphenadrine HCI*

1.168

0.005

paraquat

0.262

mercury II chloride

0.36

rat hepat, m moles

0.019

HLD, m moles

Daphnia, m moles

0.0002

Rat, m moles

hexachlorophene

Microtox, m moles

Chemical tested

IC50, m moles

1105

Rhizobium, m moles


0.002

0.0056

0.144 0.741

0.00013

0.107

0.003

673

661

52.5

906

4.52

0.0.23

0.11

3.581

0.832

0.033

0.098

0.114

2.344

0.224

0.1

0.166

1.176

0.00054

0.02

0.1

0.025

0.0022

0.098

0.05

phenol

13

0.186

3.39

4.25

0.077

1.48

0.797

potassium chloride

290

493

34.6

490

0.288

92.22

potassium cyanide

0.225

0.275

0.0776

1

0.0141

0.437

0.783

quinidine sulfate

0.422

1.202

0.617

0.036

0.0797

sodium chloride

288

562

51.3

sodium oxalate

>13

5.428

theophylline

1.926

13.8

1.349

2.63

0.724


0.583

0.342

77.6

6.71

1.48

warfarin

1.781

0.209

0.005

1.549

0.0219

0.139

xylene

1.24

0.079

34.6

0.851

0.468

17.47

pentachlorophenol

60.3

0.129 17

102 0.582

0.67

2.175

Mw = molecular weight of the compound; n = number of times compound was tested; var = variation in the results, the standard deviation /mean\ toxicity reported as mmoles toxin for the IC50 for the test. The rat test is an LD50. Rhizobium data are from this work. MicrotoxTM, HLD, and Daphnia data are from Calleja, (1993). The IC50 data (pooled data from animal cell tests) is from Halle et al., 1993. Rat hepatocyte data is from Shrivastava (1992). The IC50 samples include only 20 chemicals. The other methods involve about 34 chemicals.

In Table 15.2.2.3 the results of comparisons among these tests are summarized. The data from these 6 determinations were plotted, one assay versus another assay, the regression coefficient noted and the correlation coefficient calculated. The Rhizobium assay was the most sensitive for 12 of 33 chemicals. The Microtox assay was most sensitive for 6 of the compounds. The IC50 assay was most sensitive for 6 of the compounds. The Daphnia test was most sensitive to 10 of the compounds. The average values for the toxicity of the chemicals was lowest for the IC50, but then it was also the test with the fewest values included.

1106

James L. Botsford

Table 15.2.2.3. Comparisons for involving Rhizobium, MicrotoxTM, HLD, IC50, Daphnia magna and rats n

r2

n

r2

Rhizobium x MicrotoxTM

34

0.875

Rats x IC50

19

0.724

Rhizobium x rats

34

0.603

Rats x HLD

27

0.817

Rhizobium x IC50

20

0.906

Rats x Daphnia

32

0.685

Rhizobium x HLD

34

0.628

Rats x rat hepatocytes

34

0.748

Rhizobium x Daphnia

34

0.888

IC50 x HLD

16

0.732

Rhizobium x rat hepatocytes

34

0.764

IC50 x Daphnia

29

0.879

Microtox x rats

34

0.571

IC50 x rat hepatocytes

20

0.700

Microtox x IC50

24

0.840

HLD x rat hepatocytes

26

0.648

Microtox x HLD

31

0.728

HLD x Daphnia

29

0.702

Microtox x rat hepatocytes

33

0.700

Daphnia x rat hepatocytes

33

0.814

Microtox x Daphnia

34

0.846

n = number of samples in the test. One test was compared with another, the log of the values plotted, a regression line fit and the correlation coefficient determined from the regression coefficient, r2. The data for rats, Daphnia, HLD is from Calleja et al., 1994. The data for IC50 is from Halle et al., 1992. The data for rat hepatocytes is from Shrivastava et al., 1992.

In Table 15.2.2.4, the results of these comparisons with the results of the Rhizobium test are presented. The Rhizobium test correlates well with MicrotoxTM, Daphnia, and particularly well with the 20 samples in the IC50 test. The comparison of the Rhizobium assays with other published results has been examined in detail (Botsford, 2000a). The Rhizobium test has been compared with values from several laboratories for the MicrotoxTM assay and correlation coefficients have varied from 0.750 to 0.893 indicating the two methods provide comparable results. Four comparisons with values for Daphnia have provided correlation coefficients from 0.776 to 0.953 indicating that the Rhizobium assay agrees well with Daphnia. QSAR provides a method to predict the toxicity of a compound from the structure of the compound and the water/octanol partition coefficient (Nirmilkhandan and Speece, 1988). This method has been compared with assays using sewage sludge and PolytoxTM (Sun et al., 1993). The method compares well with both these other techniques. An examination of the data provided by tests for toxic chemicals using these techniques suggested that they are not as sensitive as more direct methods for determining toxicity. It was found that MicrotoxTM, rat hepatocytes, the Rhizobium assay, and Daphnia all provided more sensitivity to toxic chemicals than did the QSAR estimates (Botsford, 2000a). The IC50 values for these other techniques were lower than those computed using the QSAR method. QSAR data are not often found in the literature.


1107

Table 15.2.2.4. The Rhizobium assay compared with other assay methods n

cc

3T3 cells x Rhizobium

9

0.908

Ekwall et al., (1989)

Asictes cells x Rhizobium

34

0.870

Romert et al., (1994)

24

0.928

Kahru and Bordchardt (1994)

B. subtilis x Rhizobium

20

0.619

Kherzmann (1993)

Daphnia 1 x Rhizobium

35

0.891

Calleja et al., (1993)

Daphnia 2 x Rhizobium

35

0.897

Lilius et al., (1994)

pooled Daphnia x Rhizobium

14

0.775

Calleja, Liilius, Munkitrick

E. coli x Rhizobium

20

0.641

Kherzman (1993)

fathead minnow x Rhizobium

15

0.942

Munkitrick et al., (1991)

guppies x Rhizobium

9

0.950

Konemann (1981)

HepG2 cells x Rhizobium

9

0.892

Ekwali et al., (1989)

Biotox

TM

x Rhizobium

Reference

TM

x Rhizobum

35

0.803

Kaiser and Palabrica (1991)

TM

x Rhizobium

15

0.758

Munkittrick et al., (1991)

mice x Rhizobium

35

0.651

Calleja et al., (1991)

TM

1 x Rhizobium

16

0.903

Sun et al., (1993)

TM

2 x Rhizobium

15

0.796

Elanabarwy et al., (1988)

sludge x Rhizobium

16

0.853

Sun et al., (1993)

trout hepatocytes x Rhizobium

35

0.760

Lilius et al. (1994)

trout figerlingx x Rhizobium

15

0.910

Munkittrick et al., (1991)

QSAR x Rhizobium

19

0.779

Sun (1994)

pooled minnow x Rhizobium

32

0.808

Geiger et al., (1991), Munkitrick et al., (1991)

Microtox

Microtox

Polytox Polytox

n = number of samples in the comparison; cc = the correlation coefficient, the square root of the regression coefficient.

15.2.2.6 Toxic herbicides When the toxic herbicides were studied, it was found that the animal tests supplied by the manufacturers indicated that most of the herbicides were not toxic. The tests run by the manufacturers with trout fingerlings and with Daphnia indicated that most of the herbicides were quite toxic. The Rhizobium work also showed that most of the herbicides were toxic at levels lower than 1000 ppm. The trout fingerlings showed all but one of the herbicides was toxic at this level and Daphnia indicated that all but 2 were toxic (Hillaker, 1998). This shows the necessity of running “batteries of tests,” multiple tests with a compound. Every method of determining toxicity has some chemicals it cannot detect as toxic and some chemicals that are detected at very low levels. For example, the Rhizobium assay does not detect phenol as being very toxic but detects pentachlorophenol as being extremely toxic,

1108

James L. Botsford

the test is at least an order of magnitude more sensitive to pentachlorophenol than any other test found. 15.2.2.7 Toxicity of divalent cations Reduction of the dye is inhibited by divalent cations in the Rhizobium system. Common ions, calcium and magnesium, inhibit the reduction of the dye. The toxicity of the ions is shown in Table 15.2.2.5. Mercury and cadmium, generally thought to be the most toxic minerals were the most toxic with this assay. Calcium and magnesium are also toxic. Water and soil samples typically contain calcium and magnesium so in order to analyze water and soil samples for toxic organic chemicals, a method to eliminate this inhibition by metal ions was sought. Table 15.2.2.5. Toxicity of divalent cations Minerals

n

var.

m moles

Ba+2

6

33

Cd Ca

+2

+2

Co

+2

Cu

+2

12 11 11 13

41 33 46 19

Minerals

n

var.

m moles

0.109

Mg+2

9

20

0.404

0.004

+2

11

35

0.045

+2

0.05 0.009 0.007

Fe

+2

7

18

0.587

Fe

+3

14

27

0.098

Mn Hg

10

23

0.0006

+2

11

11

0.452

+2

9

39

1.849

+2

10

7

0.062

Ni Se

Zn

n = number of samples tested; var. = variation, standard deviation divided by the mean. Values reported as m moles l-1. All minerals except ferrous ion as chloride salts. Several were tested as both chlorides and sulfates with little difference. When ferrous sulfate was tested, it was made up fresh each day before the assay.

EDTA is used routinely in biochemistry to chelate divalent metal ions. It was thought this might chelate the calcium and remove it from the system. EDDA and EGTA are also used and are thought to chelate calcium more effectively than EDTA. Neither of these chelators affected the reaction, both were simply slightly toxic (about 3000 ppm). A series of experiments were run and it was found that the inhibition of reduction caused by 1.4 to 1.6 µmoles of calcium was relieved by 1 µmoles EDTA. There is not a stoichiometric relationship between EDTA and the metal ion. It is not simply chelating the metal ion. It was found that 2.5 µmoles EDTA would eliminate the toxicity of all the ions at their IC50, the concentration of cation that inhibited reduction of the dye 50%. Thus 2.5 µmoles EDTA would eliminate toxicity from 5.5 µmoles calcium but only 0.006 µmoles mercury. It was observed initially that the toxicity of most organic chemicals could be determined in the presence of 2.5 µmoles EDTA. This suggested there could be two mechanisms involved in the reduction of MTT. One is inhibited by toxic organic chemicals and the second is inhibited by divalent cations. 15.2.2.8 Toxicity of organics in the presence of EDTA A series of experiments were run looking at the toxicity of organic chemicals in the presence of 0.74 µmoles calcium (25 ppm, this concentration inhibits reduction of the dye completely) and 2.5 µmoles EDTA. Four of the 35 chemicals tested had greater toxicity with EDTA and calcium than in the controls. Two chemicals were no longer toxic. The toxicity of 16 of the chemicals was not affected by the calcium and EDTA. The toxicity of 13 chemicals was decreased by at least 10% but was not eliminated by the addition of EDTA and cal-


1109

cium. The toxicity of 4 chemicals was enhanced, was greater, when the calcium and EDTA were present. The addition of calcium and EDTA at these concentrations had no effect on the apparent toxicity of: 1,4-dinitrobenzene, the herbicide 2,4-D, 2,4-dinitrotoluene, 4-chloro-benzoate, carbon tetrachloride, chloroform, cynazine, hexachlorophene, isonicotinic acid the insecticide Lindane, o-nitrophenol, p-toluidine, the antibiotic Streptomycin, tetrachloroethylene and trichlorophenol. The addition of calcium and EDTA reduced the toxicity at least 10% of 2,6-dinitrocresol, 2,4-dinitrophenol, 2,6-dinitrophenol, 2,6-dinitrotoluene, 2-methyl resourcinol, 3-phenoxybenzoate, 2,4-dinitrocresol, the detergent CTAB, the antibiotic Neomycin, p-hydroxybenzoate, pentachlorophenol, salicylic acid, and trichloroethylene. p-amino benzoic acid and p-hydroxy benzoate were no longer toxic with these levels of calcium and EDTA. 3-chlorobenzoate, 3-methyl-4-nitro-phenol, and the detergents sodium lauryl sulfate and Tween 80 had increased toxicity with the addition of calcium and EDTA. No correlation between the structure of the chemical and the effect of EDTA and calcium could be ascertained. For example, streptomycin and neomycin are both amino glycoside antibiotics. Calcium and EDTA did not affect the toxicity of streptomycin but nearly eliminated the toxicity of neomycin. Several chemicals were tested with 25 µmoles EDTA and 7.4 µmoles of calcium, 10 times the amount used in the experiments reported. The results of this experiment are shown in Table 15.2.2.6. In the presence of high levels of EDTA and calcium, isonicotinic acid is no longer toxic. Dinitrophenol and 3-phenoxy benzoate, in the presence of low levels of EDTA and calcium were not as toxic, in the presence of high levels, it had the same toxicity. In presence of low levels of EDTA and calcium, Streptomycin was comparably toxic, the calcium and EDTA did not affect the toxicity. In the presence of high levels, Streptomycin was extremely toxic. Pentachlorophenol, a common soil contaminant, had reduced toxicity with low levels of EDTA and calcium, had elevated levels of toxicity with high levels of EDTA and calcium. This work shows that the effect of EDTA and specific toxins must be worked out before any conclusions as to the toxicity of the compound in the presence of EDTA and calcium can be established. Calcium and magnesium are commonly found in water. Obviously if this assay is to be used with water samples, EDTA must be added. It must be determined using water uncontaminated with organic toxins how much EDTA must be used to compensate for the divalent cations. Often the concentration of divalent cations is determined by atomic absorption spectroscopy. However, these values do not agree with the toxicity relieved by EDTA. Soil samples with as much as 5 gm calcium (45 mM) per kg soil have been assayed using 2.5 µmoles EDTA in each sample (Hillaker, 1996). The calcium is complexed with sulfate and phosphate ions and the calcium is not available to the cell, is not seen by the mechanism that reduces the dye. Levels of soluble calcium and magnesium in water are very low. We have found that 2.5 µmoles of EDTA relieves the inhibition caused by divalent cations in all water and soil samples tested thus far (Botsford, 2000b).

1110

James L. Botsford

Table 15.2.2.6. Effect of high concentrations of EDTR and calcium on toxicity (Inhibition in %) Compound

2.5 µmole EDTA, 0.74 µmole Calcium

25 µmole EDTA, 7.4 µmole Calcium

Control

+ EDTA, + Ca

Control

+ EDTA, + Ca

isonicotinic acid

27.0

25.4

30.9

105

2,4-dinitrophenol

44.7

63.0

58.2

50.0

Streptomycin

55.0

48.9

47.2

0.00

Neomycin

0.037

62.4

77.2

77.3

pentachlorophenol

48.0

63.7

26.0

13.3

3-phenoxybenzoate

54.3

74.0

34.5

37.8

15.2.2.9 Mechanism for reduction of the dye It is not known how the dye is reduced. It is not known why toxic chemicals inhibit the reduction. It is thought that tetrazolium dyes are reduced by cytochromes (Altman, 1975). But this has been questioned (Marshall et al., 1993). In eucaryotic cells, all the cytochromes are in the mitochondria. Marshall’s group has found that the dyes are reduced in preparations from cells with the mitochondria removed. It has been found that a mutant of Escherichia coli lacking one of two major cytochromes found in this bacterium is unable to reduce MTT (Botsford, unpublished). But in E. coli, reduction of the dye is not inhibited by toxic chemicals. This may not be an analogous situation. The dye could be reduced by a different mechanism. R. meliloti, like most other bacteria, has many reductases. Some of these are membrane associated and damage to the membrane could affect the reductase. One of these reductases could be responsible for reduction of the dye. It has been found that the MTT is transported into the cell before it is reduced. The reduced dye is inside the cells. Cells with the dye can be concentrated by centrifugation, the dye appears in the cell pellet. None of the dye is in the supernatant. Toxic chemicals could interfere with the transport of dye into the cell prior to reduction. Transposon insertion mutants unable to reduce the dye have been obtained and five mutants have been isolated. All grow very slowly in minimal media supplement with 0.1 % casamino acids and obviously all have lost a critical function. With these mutants it should be possible to clone and then to sequence the function responsible for reduction of the dye. From the sequence, the nature of the function can be determined. In our studies comparing the Rhizobium assay with other assays, it was observed that tests using viable animals were almost always less sensitive to toxins. Tests using Daphnia, the various animal cell tests and MicrotoxTM and the Rhizobium test seemed to be most sensitive. Tests using fish (fathead minnow, trout fingerlings) give results comparable to tests with the bacterial indicators. PolytoxTM and QSAR were less sensitive, had higher IC50 values, but were more sensitive than the tests using viable animals. Were the author asked to recommend a test procedure to indicate if a chemical were toxic, the author would recommend an animal cell test, probably using freshly isolated rat liver hepatocytes, the Rhizobium test and the MicrotoxTM test. Tests with freshly isolated rat hepatocytes would not require that the cells be grown in a laboratory situation and this would be much simpler. These three procedures are much more sensitive than tests involved live animals. These three tests would be simpler than tests with Daphnia. This would provide a “battery of


1111

tests”, would not offend animal rights advocates and should indicate if the chemical is dangerous. All could be performed by personnel with chemical laboratory skills. 15.2.2.10 Summary This work shows that the Rhizobium test provides results comparable to other tests. Tests seem particularly comparable to work with Daphnia magna and with results from in vitro tests with animal cells. The test is simple, unskilled laboratory workers can master it quickly. The test is inexpensive, no specialized equipment is required, given cells, any laboratory able to carry out simple chemical analysis should be able to perform the assay. The test is rapid, a sample can be tested and analyzed in an hour, the test does not take several days. It offers an ideal first test for toxic chemicals (Blaise, 1991) References Altman, F. P. 1976. Tetrazolium salts and formazans. Progress is Histochemistry and Cytochemistry 9:6-52. Bitton, G., Dutka, B. J. (1986) Toxicity testing using microorganisms. CRC Press Inc. Boca Raton, Florida 163 pp. Blaise, C. 1991. Microbiotests in aquatic toxicology. Environmental Toxicology and Water Quality. 6:145-151. Botsford, J. L., Rivera, J., Navarez, J., Riley, R., Wright T., Baker, R. 1997. Assay for toxic chemicals using bacteria. Bulletin of Environmental Contamination and Toxicology 59:1000-1008. Botsford, J. L. 1998. A simple assay for toxic chemicals using a bacterial indicator. World Journal of Microbiology and Biotechnology. 14:369-376. Botsford, J. L. 1999. A simple method for determining toxicity of chemicals using a bacterial indicator organism. Environmental Toxicology 99:285-290. Botsford, J. L. 2000a. A comparison of alternative tests for toxic chemicals. To be submitted ATLA journal. Botsford, J. L. 2000b. Role of EDTA in a simple method for determining toxicity using a bacterial indicator organism. World Journal Microbiology and Biotechnology, in press. Brodie, B., Reid, W.D., 1967. Some pharmacological consequences of species variation in rates of metabolism. Federation Proceedings 26:1062-1070. Bullich, A. A., Tung, K-K, Scheiber, G. 1990. The luminescent bacteria toxicity test: Its potential as an in vitro alternative. Journal of Bioluminesence and Chemiluminescenmce 5:71-77. Calleja, M. C., Persoone, G., Geiadi, P. (1994) Comparative acute toxicity of the first 50 multicenter evaluation of in vitro cytotoxicity chemicals to aquatic nonvertebrates. Archives Enviornmental Contamination and Toxicology 26:69-78. Calleja, M. C., Persoone, G., Gelandi P. 1993. The predictive potential of a batter of exotoxilogical tests for human acute toxicity, as evaluated with the first 50 MEIC chemicals. ATLA 21:330-349. Clemendson, C., McFarlane-Abdulla, E., Andersson, M., Barile, FA., Calleja, M.G., Chesne, C., Clotheir, R., Cottin, M. Curren, R., Dierickx, P., Ferro, M., Fiskejo G, Garza-Ocanas, L., Gomez-Lecon, M.J., Golden, M., Isomaa, B, Janus, J., Judge, P., Kahru, A., Kemp, R.B., Kerszman, G., Kristen, U. Kunimoto, M., Kaarenlapi, S., Lavrijsen, K., Lewan, L., Lilius, H., Malmsten, A., Ohno, T., Persoone, G., Pettersson, R., Roguet, R., Romert, L., Sandberg, M., Sawyer, T.W., Seibert, H., Shrivastava, R., Sjostrom, Stammati, A., Tanaka, N., Torres-Alanis, O., Voss, J-U. Wakuri,S., Walum, E., Wang, X., Zucco, F., Ekwall, B. (1996). MEIC evaluation of acute systemic toxicity. ATLA 24:273-311. Ekwall, B, Bondesson, 1, Catell, J.V., Gomez-Lechon, M. J., Heiberg, S., Hogberg, J. Jover, R., Ponsoda, X., Rommert, L., Stenberg, KL., Walum, E. (1989) Cytoxocity evaluation of the first ten MEIC chemicals: Acute lethal toxicity in man predicted by cytotoxicity in five cellular assays and by oral LD50 tests in rodents. ATLA 17:83-100. Ekwall, B., Johansson, A. 1980. Preliminary studies on the validity of in vitro measurements of drug toxicity using HeLa cells I. Comparative in vitro cytotoxicity of 27 drugs. Toxicology Letters 5:299-307. Einabarawy, M. T., Robideau, R. R., Beach, S. A. (1988) Comparison of three rapid toxicity test procedures: MicrotoxTM, PolytoxTM and activate sludge respiration inhibition. Toxicity Assessment 3:361-370. Fresjog, G. 1985. The allium test as a standard in environmental monitoring. Hereditas, 102:99-112. Gaggi, C., Sbrilli G., A.M. Hasab El Naby, Bucci, M., Duccini, M., and Bacci, E. 1994. Toxicity and hazard raking of S-triazine herbicides using Microtox, 2 green algal systems and a marine crustacean. Environmental Toxicology and Chemistry 14:1065-1069.

1112

James L. Botsford

Geiger, D. L., Brooke, L. T., Call, D. J. editors (1990) Acute toxicities of organic chemicals to fathead minnows (Pimphates promeias). Center for Lake Superior Environmental Studies. University of Wisconsin, Superior, Wisconsin. 900 pp. Gupta, G., Karuppiah, M. 1996. Toxicity identification of Pocomoke River porewater. Chemosphere 33:939-960. Halle, W., Baeger, I., Ekwall, B., Spielmann, H. 1991. Correlation between in vitro cytotoxicity and octanol/water partition coefficient of 29 substances from the MEIC program. ATLA 19:338-343. Hillaker, T. L. 1996. An assay for toxic chemicals using Rhizobium method as the indicator: Use of this test with agricultural herbicides. MS Thesis, Biology. New Mexico State University. Jaffe, R. L., 1995. Rapid assay of cytotoxicity, using Teratamitus flagellates. Toxicology and Industrial Health 11:543-553. Jung, K., Bitton, G. 1997. Use of CeriofastTM for monitoring the toxicity of industrial effluents: Comparison with the 48-H acute Ceriodaphnia toxicity test and MicrotoxTM. Experimental Toxicology and Chemistry 16:2264-2267. Kahru, A., Borchardt, B. 1994. Toxicity of 39 MEIC chemicals to Bioluminescent photobacteria (The BiotoxTM test): Correlation with other test systems. ATLA 22:147160. Kaiser, K.L.E., Palabrica, V. S., 1991. Photobacterium phosphoreum toxicity data index. Water Pollution Research Journal of Canada. 26:361-431. Kerszman, G. 1993. Toxicity of the first ten MEIC chemicals to bacteria. ATLA 21:151155. Kerzman, G. 1993. Of bacteria and men: Toxicity of 30 MEIC chemicals to bacteria and humans. ATLA 21:233-238. Konemann, H, 1981, Quantitative structure-activity relationship in fish toxicity studies. Toxiocology 19:209-221. Lilius, H., Isomaa, B., Holstrom, T. A comparison of the toxicity of 50 reference chemicals to freshly isolated rainbow trout hepatocytes and Daphnia magna. Aquatic Toxicology 30:47-60. Liu, D. 1989. A rapid and simple biochemical test for direct determination of chemical toxicity. Toxicity Assessment 4:389-404. Marshall, N. J., Goodwin, C. J., Holt, S. J. 1995. A critical assessment of the use of microculture tetrazolium assays to measure cell growth and function. Growth Regulation 5:69-84. Mossman, T . 1983. Rapid colorimetric assay for cellular growth and survival: Application to proliferation and cytotoxicity assays. Journal of Immunological Methods 65:55-63. Munkittrick K. R., Power, E. A., Sergy, G. A. 1991. The relative sensitivity of MicrotoxTM, Daphnia, rainbow trout, fathead minnow acute lethality tests. Environmental Toxicology and Water Qualtity 6:35-62. Nirmalakhandan, N. N., Speece, R. E. 1988. Prediction of aqueous solubility of organic chemicals based on molecular structure. Environmnental Science and Technology. 22:328-338. Robertson, B. 1996. Developing a technique to lyophilize Rhizobium mefloti MS thesis, Biology, New Mexico State University. Rodericks, J. V.1992. Calculated Risks. Cambridge, Cabridge University Press 256 pp Romert, L., Jansson, T. Jenssen, D. 1994. The cytotoxicity of 50 chemicals from the NEIC study determined by growth inhibition of Ascites Sarcoma BP8 cells: A comparison with acute toxicity data in man and rodents. Toxicology Letters 71:39-46. Rouguet, R., Cotovia, J. Gaetani, Q., Dossou K. G. Rougier, A. 1993. Cytotoxicity of 28 MEIC chemicals to rat hepatocytes using two viability endpoints: correlation with acute toxicity data in rat and man. ATLA 1:216-224. Ruelius, H. W. 1987. Extrapolation from animals to man: predictions, pitfalls and perspectives. Xenobiotica 17:255-265. Shrivastava, R., Deiominie, C., Chevalier, A., John, G., Ekwall, B., Walum, E. Massingham, R. Comparison of in vitro acute lethal potency and in vitro cytotoxicity of 48 chemicals. Cell Biology and Toxicology 8:157-167. Snell, T. W., Personne, G. 1989. Acute bioassays using rotifers. II. A freshwater test with Brachionus rubens. Aquatic Toxicology 14:81-92. Stephenson, G. L., Kausik, N. K., Solomon, K. R. 1991. Chronic toxicity of a pure and technical grade pentachlorophenol to Daphnia magna. Archives Environmental Contamination and Toxicology. 21:388-394 Sun, B., Nimalakhandan, N., Hall, E., Wang, X. H., Prakash, J., Maynes, R. (1994) Estimating toxicity of organic chemicals to activated-sludge microorganisms. Journal of Environmental Engineering 120:1459-1469. Sun, B. (1993). Comparison of interspecies toxicity of organic chemicals and evaluation of QSAR approaches in toxicity prediction. MS Thesis, Environmental Engineering. New Mexico State University. Thomulka, K. W., McGee, D. J., Lange, J. H. 1993. Detection of biohazzardous materials in water by measuring bioluminescence with the marine organism Vibrio harveyi. Journal Environmental Science and Health. A28: 2153-2166. Toussaint, M. W., Shedd, RT. R., van der Schalie, W. H., Leather, G. R. (1992) A comparison of standard acute toxicity tests with rapid-screening toxicity tests. Environmental Toxicology and Chemistry 14:907-915.

15.2.3 An innovative GC method

1113

15.2.3 DESCRIPTION OF AN INNOVATIVE GC METHOD TO ASSESS THE INFLUENCE OF CRYSTAL TEXTURE AND DRYING CONDITIONS ON RESIDUAL SOLVENT CONTENT IN PHARMACEUTICAL PRODUCTS

Christine Barthélémy Laboratoire de Pharmacie Galénique et Biopharmacie Faculté des Sciences Pharmaceutiques et Biologiques Université de Lille II, Lille, France

Michel Bauer International Analytical Sciences Department Sanofi-Synthélabo Recherche, Toulouse, France

The presence of residual solvents (RS) in pharmaceutical substances occurs for various reasons. Solvents are involved in all steps of raw material synthesis and pharmaceutical productions. The search for the presence of RS in a pharmaceutical product and their concentrations are now mandatory in any new monographs (as detailed in Chapter 16.2). The RS remaining in the crystals of pharmaceutical products may be the cause of health disorders because, when a drug is taken every day, chronic toxicity may occur. The presence of RS may have other consequences, such as modifying stability, organoleptic characters, pharmacotechnical parameters (flow properties, crystalline form, compression ability) and biopharmaceutical characteristics, that may fluctuate according to RS content (as detailed in chapters 14.21.1 and 16.2). It is therefore necessary to reduce the residual solvents contained in crystalline particles as much as possible. It is well known that solvents can exist in three different states within the crystals: • Solvents adsorbed on the crystal faces: these are generally easily desorbed during conventional drying because the binding forces between solvents and crystals are very weak. • Occluded solvents such as microdroplets in the crystal: these are often difficult to extract. Generally, they can escape when the crystal is being dissociated: - during grinding, potentially leading to clodding, - during storage leading to very compact aggregates. • Solvents bound to drug molecules in the crystal and known as “solvates”. These bound solvents escape at a characteristic temperature, producing desolvated forms; the solvate and the desolvated forms are two different crystalline entities that can exhibit very different mechanical behaviors. The main objective of any chemist crystallizing pharmaceutical raw materials should be either the total elimination of the organic solvents or the significant reduction of RS level in order to be below the regulatory limits.1 15.2.3.1 Description of the RS determination method There are several analytical methods to assess the RS content of drugs. Among them, the gas chromatography (GC) is largely preferred. We refer to the chapter 16.2 for more details. In direct injection methods, the products in which RS are included are usually dissolved in an appropriate solvent and then directly injected into a gas chromatograph (GC). The main problem with these techniques is that non-volatile substances are gradually retained in the column, causing a rapid decrease in its sensitivity and efficiency; this is one of the reasons why headspace techniques are increasingly used instead.

1114

Christine Barthélémy, Michel Bauer

To avoid these problems, we have developed a method consisting initially of the complete dissolution of the substance in an appropriate solvent followed by micro-distillation and then finally of the injection of the distillate into the GC. This technique allows for the complete recovery of RS without the drawbacks mentioned above.2,3 The operating conditions are as follows: the powders to be analyzed are poured into a micro-distilling flask. An appropriate solvent is chosen to allow the product as well as the RS to be completely dissolved. The solvent Figure 15.2.3.1. Photograph of the micro-distilling deshould not interfere with the RS extracted; for vice. example, in the following studied cases: methanol for dioxane, chloroform, tetrahydrofuran, hexane and dichloromethane determinations and 1-butanol for ethanol determination. Assay solutions were obtained by completely dissolving 500 mg of the product to be tested in 5 ml of the appropriate solvent. The micro-distillation is carried out until complete dryness and is followed by condensation of the solvent vapor. The distillate is collected in a gauge glass set on ice. The assembly of the system is photographed on Figure 15.2.3.1. In this way the non-volatile substances stay entrapped in the micro-distilling flask and the totality of the RS are recovered. The distillate is then directly injected into the GC. The method was validated (specificity, linearity, repeatability, reproducibility...) by distilling, in the same operating conditions, defined standard solutions of solvents. The recovery of RS is very good as can be seen in the example reported on Table 15.2.3.1. 15.2.3.2 Application: Influence of crystal texture and drying conditions on RS content After washing the crystals, drying is the most effective way of lowering the content of organic volatile impurities. The required drying conditions differ greatly according to the solvent state in the crystalline particles, to the thermodynamic events that can occur when the substance is heated, and also to the texture of these particles. Table 15.2.3.1. Example of a calibration curve and recovery calculations of dioxane distilled in methanol Dioxane standards area under curve and variation coefficient (%)

Dioxane recovered after distillation area under curve and variation coefficient (%)

% and ppm recovered

44.8 ppm

17161.0± 705.5, v.c: 4.11 % 16819.3± 631.5, v.c: 3.75 % 98.01 %, 43.9 ppm

56 ppm

25904.7±1493.1, v.c: 5.76 %

25869.0± 748.4, v.c: 2.89 % 99.86 %, 55.9 ppm

112 ppm

66147.3±1010.4, v.c: 1.53 %

66831.7± 472.6, v.c: 0.71 % 101.03 %, 113.2 ppm

168 ppm

108157.7±2026.6, v.c: 1.53 %

107361.0±1546.4, v.c: 1.44 %

99.26 %, 166.8 ppm

224 ppm

146674.0±1490.4, v.c: 1.02 %

146722.7±2469.7, v.c: 1.68 %

100.03 %, 224.1 ppm

linear regression

Y = 724.74 X - 14855.6

Y = 724.29 X - 14889.7

correlation coeff.

0.9999

0.9999

mean : 99.64± 1.11 % v.c : 1.1 %


1115

Among all parameters influencing RS content, crystal texture is of utmost importance. It is evident that a crystal exhibiting a porous texture will enable the easy escape of a solvent while a compact and dense crystal will retain the solvent inside its structure whatever the type of particles: monocrystalline (i.e., monoparticular) or “polycrystalline”. The term “polycrystalline” particles will be employed to designate elementary particles that can be composed of agglomerates, spherolites or “spherical crystals” according to Kawashima.4 To illustrate the importance of the texture of particles on RS content, we can consider some examples taken from our laboratory experiments. Several crystallization processes were investigated on pharmaceutical products with different solvents leading either to compact monocrystalline particles or “sintered-like” (i.e., microcrystallites fitted into each other and partially welded involving a porous texture) or polycrystalline particles which were more or less dense.2,3 The physical study of these particles such as, optical microscopy, particle size analysis, electron scanning microscopy and thermal analysis, have been used to link the RS contents with the drying conditions of crystals and to demonstrate that the optimal drying conditions differ greatly according to the texture of the particles. As we shall see in a first example of monocrystalline particles of paracetamol recrystallized in dioxane, a wide open texture is generally favorable to a low RS content after a progressive drying at a moderate temperature to avoid the formation of a superficial crust. In fact, in this case, the drying conditions of crystals highly influence their residual solvent content. Optimal drying conditions seem to be the progressive and moderate ones. In contrast, too drastic drying conditions may hinder the solvent escape by a “crust” effect. This crust is due to a drying temperature which is too high, leading to a melt and a dissolution of the surface of the crystals. Finally, when a desolvation occurs during the drying, it can modify the texture of crystals and form a crust. The intensity of the phenomenon depends on the solvent. To remove the solvents efficiently, it is then necessary to exceed the desolvation temperature of the solvates that may be produced during crystallization. With this example we can clearly point out that the crystal texture is a determining factor in the complete escape of the solvent: when the crystalline texture is sintered-like, after the desolvation of solvate crystals, progressive drying is necessary to prevent the “crusting” phenomenon. The drying conditions should also be adapted to the area offered to the evaporation of solvent, particularly in the case of polycrystalline particles presenting a high porosity and a large surface to be dried. This important surface is due to the disordered rearrangement of very small crystals inside the particles. To illustrate this, we shall see in a second example the cases of spherical crystals of meprobamate and ibuprofen agglomerates. In both cases the RS can escape easily from the large surface of polycrystalline particles. Lastly, we shall see in a third example, that we must not forget that thermodynamic phenomenon can occur under drying. This is a very particular example because the pharmaceutical product used (paracetamol) presents a polymorphic transition that can occur during the drying phase and lead to a new organization inside the crystal, allowing the escape of the RS. This can be very interesting in the case of products with a low transition temperature. The usual polymorphic form of commercialized paracetamol is the monoclinic form; but in particular cases small amount of metastable orthorhombic form can be obtained. In the case of paracetamol, at the transition temperature (156°C) we observe a brutal solvent escape that could be due to the solid-solid transition undergone by the orthorhombic form into the monoclinic one. The disorder produced and the increase of the specific volume occurring during this first order transition allow the occluded solvents to escape more easily.

1116


15.2.3.2.1 First example: monocrystalline particles of paracetamol Preparation of monocrystalline particles In this first example, paracetamol was recrystallized separately in three different types of solvents (ethanol, water and dioxane) with different boiling points, molecular weights, dielectric constants and paracetamol solubilising power. After the crystallization process,5 crystals were separated by filtration under vacuum and washed with the same crystallization solvent. Each batch was divided into four fractions to be dried differently. Particle drying conditions The paracetamol crystals were submitted to four different drying conditions2 (Table 15.2.3.2): Drastic: under vacuum at 100°C for 3 hours. Drastic: in a ventilated oven at 100°C for 3 hours. Progressive: in a ventilated oven at 60°C for 1 hour, at 80°C for 1 hour and finally at 100°C for 1 hour. Very moderate: under ventilated hood at 20°C for 24 hours and then in a ventilated oven at 35°C for 48 hours. Morphological aspect of the particles Dried crystals were observed by optical and scanning electron microscopy (SEM) in order to measure their Ferret mean diameter and to determine their habit and texture. The mean diameters of crystals obtained from different solvents and submitted to different drying conditions are reported in Table 15.2.3.3. Table 15.2.3.2. Drying conditions studied on the different particles Type of particles

monocrystalline

Drugs

paracetamol

Type of drying

Temperature

Time

drastic under vacuum drastic under ventilation

100°C 100°C 60°C + 80°C + 100°C 20°C + 35°C 90°C 60°C + 75°C + 90°C 60°C 40°C 156°C 156°C 100°C 100°C + 156°C 60°C + 80°C + 100°C 60°C + 80°C + 100°C + 156°C

3h 3h 1h +1h +1h 24 h + 48 h 2h 30 min + 30min +1h 2h 2h 10 min 30 min 2h 2h + 10 min 30 min + 30 min +1h 30 min + 30 min +1h + 10 min

progressive very moderate drastic

meprobamate

ibuprofen

polycrystalline

progressive moderate very moderate drastic flash drastic flash drastic drastic + flash

paracetamol progressive

progressive + flash


Figure 15.2.3.2. SEM photograph of paracetamol recrystallized in dioxane and submitted to moderate drying (Photograph from reference2).

1117

Figure 15.2.3.3. SEM photograph of paracetamol recrystallized in dioxane and submitted to progressive drying (Photograph from reference2).

The optical microscopy investigation shows some significant differences in mean diameter and crystal habit due to crystallization conditions. In the case of water and ethanol, the crystals are transparent and rather regular, their habit is generally parallelepipedal. The only difference to be noted is the mean diameter of these particles; the particles recrystallized in water being 3 times larger. ESM reveals a surface that remains smooth in all drying conditions; their texture seems compact and dense. In contrast, crystals obtained from dioxane are rather different to those obtained from ethanol and water: they are opaque to transmitted light and their habit is irregular. Figure 15.2.3.4. SEM photograph of paracetamol Their mean diameter is slightly smaller than recrystallized in dioxane and submitted to drastic drying crystals from ethanol. (Photograph from reference2). It is to be noted that all the crystals obtained from the three solvents present the same polymorphic form: the monoclinic form. Table 15.2.3.3: Mean Ferret diameters of particles (Data from references2,3) Monocrystalline particles paracetamol / dioxane paracetamol / ethanol paracetamol / water

Mean diameter, µm 122 149 538

Polycrystalline particles meprobamate spherical crystals ibuprofen agglomerates paracetamol spherolites

Mean diameter, µm 145 350 113

1118


Scanning electron microscopy (SEM) carried out on these crystals reveals different behaviors according to the drying conditions tested. The crystals obtained from dioxane are similar to blocks of sintered particles since they consist of microcrystallites bound to each other, as if partially welded.5 In fact, the measurement of mercury porosity indicated a very high porosity. This sintered aspect concords with the non-transmission of light through the whole crystal. In fact, paracetamol forms a solvate with dioxane.5 Its departure does not modify the crystal habit but the surface can be seen as perforated. When the drying is very moderate, the crystal surface is perforated by numerous holes through which solvent escapes; these anfractuosities give the impression of a porous sintered-like texture (Figure 15.2.3.2). With progressive drying, the number of holes decreases (Figure 15.2.3.3). On the contrary, when the drying is drastic, the surface is relatively smooth (Figure 15.2.3.4). This can be explained by the too drastic drying conditions involving the formation of a crust at the surface of the crystal preventing the escape of the solvent from the crystal. Residual solvent determination Gas Phase Chromatography (for dioxane and ethanol) was performed on a Varian 1440 Chromatograph with a Flame Ionization Detector; packed column Porapack Super Q (Alltech, France), mesh range 80/100, length 1.8 m; internal diameter 2.16 mm; carrier gas was nitrogen (40 ml/min); injector: 210°C; detector: 250°C. For dioxane: column temperature: isotherm at 170°C. Injection: 10µl. Retention Times (RT): methanol 1.5 min; dioxane 16 min. For ethanol: column temperature: isotherm at 210°C. Injection: 5µl. RT: ethanol 1.5 min; 1-butanol 4 min. Determination of residual water on paracetamol crystallized in water: according to the titrimetric direct method of Karl Fischer. The residual solvent concentrations of the crystals obtained are reported in Table 15.2.3.4. Table 15.2.3.4. Residual solvent content of crystals obtained from different solvents and submitted to different drying conditions (Data from reference2) Drastic

Drastic under vacuum

Progressive

Moderate

Dioxane

126 ppm

183 ppm

25 ppm

53 ppm

Ethanol

2045 ppm

2501 ppm

1495 ppm

2072 ppm

0.59 %

0.68 %

0.51 %

0.56 %

Water

Whatever the drying conditions may be, the content of dioxane is always lower than the 380 ppm ICH limit1,6 (Table 15.2.3.5). Progressive drying always gives lower level residual solvent. In fact, drastic drying, in a ventilated oven and under vacuum, leads to the formation of a superficial “crust” which hinders the solvent escape; this can be clearly visualized on the scanning electron photomicrograph (Figure 15.2.3.4); and can explain the relatively high content of residual solvents.


1119

Table 15.2.3.5. Solvent class and concentration limits in pharmaceutical products (Data from references1,6) Class

Concentration limit, ppm

Chloroform

2

60

Dioxane

2

Ethanol

3

Class

Concentration limit, ppm

Hexane

2

290

380

Dichloromethane

2

600

5000

Tetrahydrofuran

3

5000

15.2.3.2.2 Second example: polycrystalline particles of meprobamate and ibuprofen Preparation of polycrystalline particles The polycrystalline particles were produced using various crystallization processes and designed to obtain directly compressible particles of pure drug, as tablets cannot be formed by direct compression of the raw materials. Spherical crystals of meprobamate These spherical particles were prepared following the usual preparation process for spherical crystals described by Guillaume.7 Spherical crystals of meprobamate appear when stirring a mixture of three liquids in the crystallizer: methanol allowing meprobamate to dissolve; water, as a non-solvent, causing meprobamate precipitation; chloroform as a bridging liquid to gather in its dispersed droplets, meprobamate microcrystallites that finally form “spherical crystals”. Ibuprofen agglomerates Ibuprofen agglomerates were prepared by a phase separation process in a mixture of ethanol and water (50/50 v/v).8 The saturated solution obtained at 60°C was constantly stirred and cooled down to room temperature. Particle drying conditions Depending on the melting point of the drugs, different drying conditions were applied in a ventilated oven. Drastic and progressive temperature conditions were studied for each type of polycrystalline particles (Table 15.2.3.2). Spherical crystals of meprobamate Taking into account the melting point of meprobamate (105°C), temperatures higher than 90°C must be avoided. The different drying conditions were: drastic drying at 90°C for 2 hours and progressive drying at 60°C for 30 minutes, then at 75°C for 30 minutes and finally at 90°C for one hour. Ibuprofen agglomerates This is a particular case; a low drying temperature must be applied because of the very low melting point of ibuprofen (76°C). The two drying conditions studied were: 40°C for two hours or 60°C for two hours. Morphological aspect of the particles Spherical crystals of meprobamate Meprobamate crystals appear as more or less rounded opaque particles (Table 15.2.3.3). Their consistence is friable. Particle size distribution is very narrow. SEM photograph of meprobamate crystals shows nearly spherical particles; their surface seems apparently smooth (Figure 15.2.3.5). However, at high magnification, the surface appears to be coated with flat crystals (Figure 15.2.3.6).

1120

Figure 15.2.3.5. SEM photograph of “spherical crystals” of meprobamate (Photograph from reference3).

Figure 15.2.3.7. SEM photograph of the cross-section of “spherical crystals” of meprobamate (Photograph from reference3).


Figure 15.2.3.6. SEM photograph of the surface (high magnification) of “spherical crystals” of meprobamate (Photograph from reference3).

Figure 15.2.3.8. SEM photograph of ibuprofen agglomerates (Photograph from reference3).

The cross-section surface shows a dense tangling-up of flat crystals inside the particle (Figure 15.2.3.7). Flat crystals are concentrically disposed at the periphery of the rounded particles, like a shell. The inside of the particle is a disordered arrangement of small flat crystals. Thus, the surface offered to evaporation is very important and this should lead to the easy escape of RS. Ibuprofen agglomerates Ibuprofen particles are large agglomerates of flat crystals with quite soft consistence and a nearly rounded shape (Table 15.2.3.3).


1121

Ibuprofen crystals observed under SEM are in total disorder (Figure 15.2.3.8). This disordered arrangement of microcrystallites inside the polycrystalline particles gives them a high isotropy of particles arrangement that should improve their compression capacity. Residual solvent determination For organic solvents, gas phase chromatography was performed on a Shimadzu GC-14B chromatograph fitted with a Flame Ionization Detector and a CR-6A Shimadzu integrator. The packed column was Porapack super Q (Alltech, France), mesh range 80/100, length 1.80 m, internal diameter 2.16 mm. Carrier gas: anhydrous nitrogen. Injector temperature: 200°C. Detector temperature: 220°C. The chromatographic conditions were: • For chloroform in meprobamate crystals: isotherm at 150°C, injection: 5 µl, RT: methanol 0.8 min, chloroform 6.7 min. • For ethanol in ibuprofen crystals: isotherm at 170°C, injection: 5 µl, RT: ethanol 1.1 min, 1-butanol 5.3 min. Determination of residual water was carried out using Karl Fischer’s titrimetric direct method after calibration with natrium tartrate and dissolution of ibuprofen or meprobamate crystals in methanol. The residual solvent concentrations of the polycrystalline particles are reported in Tables 15.2.3.6 and 15.2.3.7. Table 15.2.3.6. Residual solvent concentration of meprobamate spherical crystals submitted to different drying conditions (Data from reference3) Chloroform

Water

Progressive drying*

345±22 ppm

0.71%±0.07%

Drastic drying**

321±8 ppm

0.21%±0.01%

*30 min at 60°C + 30 min at 75°C + 1 hour at 90°C; **2 hours at 90°C

Table 15.2.3.7. Residual solvent concentration of ibuprofen agglomerates submitted to different drying conditions (Data from reference3) Ethanol

Water

2 hours at 40°C

42±3 ppm

0.76%±0.12%

2 hours at 60°C

21±4 ppm

0.44%±0.06%

As far as meprobamate spherical crystals are concerned, no significant differences are to be observed between drastic and progressive drying. No crusting phenomenon appears on meprobamate spherical crystals due to the loose tangling up of crystals, and so the solvent may escape easily between them. The residual ethanol content of ibuprofen agglomerates is very low because of the open texture of agglomerates. Moreover, the crystallization phenomenon was relatively slow, enabling the solvent to escape from crystals in formation. The higher the temperature, the lower the residual ethanol content. Both these crystals have a very porous texture. It seems that progressive drying is not essential as far as the polycrystalline particles are concerned. If we consider the official limits1,6 reported on Table 15.2.3.5 for residual solvent contents, we can note that the concentration of chloroform in meprobamate spherical crystals is much higher than the limit allowed in any drying conditions. Due to its inherent toxicity, this solvent should be avoided in the recrystallization process of meprobamate. The solvent

1122


content obtained for ethanol in ibuprofen agglomerates is very low, in all drying conditions; in all cases it is considerably lower than the tolerated limits. 15.2.3.2.3 Third example: polycrystalline particles of paracetamol Preparation of polycrystalline particles Paracetamol agglomerates were prepared by the spontaneous precipitation of paracetamol into a mixture containing hexane, tetrahydrofuran and dichloromethane9 under stirring. All the crystals obtained were filtered under vacuum. Particle drying conditions The melting point of paracetamol being 169°C and the transition temperature being 156°C, different drying conditions were tested (Table 15.2.3.2): • Drastic flash drying: 156°C for either 10 or 30 minutes, • Drastic drying: 100°C for 2 hours, • Progressive drying: 60°C for 30 minutes, then 80°C for 30 minutes, and finally, 100°C for 1 hour. After drastic drying at 100°C and progressive drying, a test with complementary drying at 156°C was carried out for 10 minutes. All the recovered crystals were packaged in glass flasks before gas phase chromatography and other analysis. Morphological aspect of particles Several types of texture and morphology are to be observed in polycrystalline particles, according to solvent proportions. The most interesting and particular example are spherical polycrystalline particles which have a radial texture (spherolites) and appear as urchin-like particles (Table 15.2.3.3). SEM reveals the very typical surface crystallization of the agglomerates (Figure 15.2.3.9). They are made up of parallelepipedal flat crystals arranged perpendicularly to a central nucleus and they are relatively strong. The implantation of peripherical crystallites is perpendicular to the surface. As it has been clearly demonstrated by Ettabia10 a Figure 15.2.3.9. SEM photograph of paracetamol ag- nucleus is formed first and then, in a second glomerates (Photograph from reference3). step, microcrystallites grow on it. Residual solvent determination For organic solvents, gas phase chromatography was performed on a Shimadzu GC-14B chromatograph fitted with a Flame Ionization Detector and a CR-6A Shimadzu integrator. The packed column was Porapack super Q (Alltech, France), mesh range 80/100, length 1.80 m, internal diameter 2.16 mm. Carrier gas: anhydrous nitrogen. Injector temperature: 200°C. Detector temperature: 220°C. The chromatographic conditions were: For dichloromethane, tetrahydrofuran and hexane in paracetamol crystals: isotherm at 150°C, injection: 5 µl, RT: methanol 0.8 min, dichloromethane 3 min, tetrahydrofuran 7.8 min, hexane 11.8 min.


1123

The residual solvent concentrations of the polycrystalline particles are reported in Table 15.2.3.8. The wide-open texture of paracetamol spherolites hinders the crusting phenomenon; it is therefore normal that there should be no difference between the effectiveness of the two types of drying: progressive and drastic (at 100°C for two hours). However, a surprisingly good result is observed when the drying temperature is 156°C whereas a prolonged time at 100°C does not improve the solvent escape in spite of the low boiling points of solvents, all much lower than 100°C (dichloromethane: 39.5°C; tetrahydrofuran: 66°C; hexane: 69°C). Using Figure 15.2.3.10. DSC and thermogravimetry curves of a DSC method, it can be observed that a paracetamol agglomerates. slight non constant exotherm at 85°C followed by a small constant endotherm at nearly 156°C occurs before the melting endotherm at 169°C (Figure 15.2.3.10). In fact, the paracetamol obtained by the crystallization process described is mainly the usual monoclinic form. However, as has been demonstrated by Ettabia10 the formation of a certain amount of the amorphous form, causes the slight exotherm. The amorphous form recrystallized at 85°C into the orthorhombic metastable form, which transits into monoclinic form at 156°C, causing the small endotherm observed at nearly 156°C. The thermogravimetric analysis shows that no solvation occurs during crystallization. Furthermore, the sudden solvent escape is not specific to one definite solvent, whereas all the contained solvent amounts dramatically decrease at 156°C. The loss of mass is high (about 0.7%) when the substance is about to melt (Figure 15.2.3.10). Table 15.2.3.8. Residual solvent concentration of paracetamol agglomerates submitted to different drying conditions (Data from reference3) "Flash drying" 156°C for 10 min for 30 min PD* PD*+ “flash” (10 min at 156°C) DD ** DD**+"flash" (10 min at 156°C)

Dichloromethane

Tetrahydrofuran

Hexane

< 200 ppm < 200 ppm 315±3 ppm < 200 ppm 309±3 ppm < 200 ppm

1095±5 ppm 630±24 ppm 2066±1 ppm 883±10 ppm 2014±30 ppm 873±2 ppm

167±1 ppm 88±4 ppm 305±3 ppm 145±5 ppm 313±4 ppm 140±1 ppm

*

PD: Progressive drying (30 min at 60°C + 30 min at 80°C + 1 h at 100°C), **DD: Drastic drying (2 h at 100°C)

Paracetamol, containing three different solvents, exhibits low dichloromethane content (200 to 320 ppm). The referencing limits remain higher than the experimental results when these particles are submitted to different drying conditions (Table 15.2.3.5). The same observation is valid for tetrahydrofuran for which the allowed upper limit is 5000 ppm. But the residue in hexane is sometimes above the regulatory threshold. The only way to be definitely below this limit is to heat paracetamol to 156°C; this confirms the advantage of flash drying this substance. As previously described, this temperature is critical for paracetamol recrystallization in the solvents used.

1124


This study enables us to underline some considerations about the drying conditions of crystals. It is important to note that not only should the usual parameters be taken into account, such as the melting point of the substances and the boiling point of the solvents, but also the texture of the particles and the thermodynamic events that can occur when the substance is heated. The kinetic of temperature increase must be chosen with respect to crystalline particle texture to obtain optimal drying conditions for lowest RS content. Progressive drying can give better solvent elimination when the texture is very finely sintered to avoid the crusting effect, which hinders any ulterior solvent escape.2 When a solvate is formed, the drying temperature must be either equal to or higher than the desolvation temperature. Lastly, the knowledge of the existence of solid-solid transitions could be in this respect of great interest. Thermogravimetry and DSC analysis are, among others, very efficient tools to assess these phenomena. References 1 2 3 4 5 6 7 8 9 10

European Pharmacopoeia, 3rd edition, addendum 1999, pp. 216-224. C. Barthélémy, P. Di Martino, A-M. Guyot-Hermann, Die Pharmazie, 50, 609 (1995). A. Ettabia, C. Barthélémy, M. Jbilou, A-M. Guyot-Hermann, Die Pharmazie, 53, 565 (1998). Y. Kawashima, M. Okumura, H. Takenaka, Science, 216, 1127 (1982). J-M. Fachaux, A-M. Guyot-Hermann, J-C. Guyot, P. Conflant, M. Drache, S. Veesler, R. Boistelle, Powder Techn., 82, 2, 123 (1995). Note for Guidance on impurities: Residual solvents, Drugs made in Germany, 41, 98 (1998). F. Guillaume, A-M. Guyot-Hermann, Il farmaco, 48, 473 (1993). M. Jbilou, A. Ettabia, A-M. Guyot-Hermann, J-C. Guyot , Drug Dev. & Ind. Pharm., 25, 3, 297 (1999). A. Ettabia, E. Joiris, A-M. Guyot-Hermann, J-C. Guyot, Pharm. Ind., 59, 625 (1997). A. Ettabia, European Thesis, Lille II University, France (8/10/1997).

16

Residual Solvents in Products 16.1 RESIDUAL SOLVENTS IN VARIOUS PRODUCTS George Wypych ChemTec Laboratories, Inc., Toronto, Canada

There are physical and chemical barriers that control solvent removal from solid-solvent systems. The most basic relation is given by the following equation: W =

P1 Kw

[16.1.1]

where: W P1 Kw

equilibrium fraction of residual solvent partial pressure of solvent in vapor phase Henry’s law constant

Both the partial pressure and Henry’s law constant depend on temperature, pressure, and solvent properties. This relationship does not consider interaction between solute and solvent. In the case of polymers, the Flory-Huggins theory gives a simplified relationship for low concentrations of solvent: ln

P1

= ln φ1 + 1 + χ

P10

[16.1.2]

where: P10 φ1 χ

vapor pressure of pure solvent volume fraction of solvent Flory-Huggins interaction parameter

Vapor pressures of some solvents can be found in the referenced monograph.1 The weight fraction of residual solvent at equilibrium can be calculated from the following equation, which accounts for polymer-solvent interaction: W =

P1 ρ 1 exp− (1 + χ ) P10 ρ 2

where: ρ1 ρ2

density of solvent density of polymer

[16.1.3]

1126

George Wypych

The last equation does not give the real values of residual solvents because equilibrium is not attained in real drying processes 0.8 and the prediction of different interactions by the interaction parameter is too simplistic. The 0.6 real values are substantially higher and the real barriers of solvent 0.4 removal more complex. These are discussed below. In real systems, several phenomena take place. These include 0.2 chemical interaction between the functional groups of polymer and the solvent. These are mostly re0 0 2 4 6 8 10 lated to hydrogen bond formation. The crystalline structure of polyTime, min mer is responsible for the modifiFigure 16.1.1. Number of n-heptane molecules per one mer of poly- cation of the diffusion process. Solvent properties determine difstyrene vs. drying time. [Data from L A Errede, P J Henrich, fusion. Polymer properties are reJ N Schrolpfer, J. Appl. Polym. Sci., 54, 649 (1994).] sponsible for the macromechanism of solvent removal from a highly viscous polymer. In addition, some real systems make use of stripping solvents which are designed to help in removal of trace quantities of solvents by use of stripping solvent displacing process solvent. These various factors interplay and determine the result. Figure 16.1.1 shows the number of residual solvent molecules per one mer of polystyrene. It is evident that solvent removal has zero-kinetics until its concentration is decreased to about 0.2 molecules of solvent per mer. It is also true that some solvent remains after drying. Even after 24 h drying, 0.06% solvent remains. These data indicate that there is a different mechanism of removing residual solvent. It is not clear if this is because of interactions, a change in the glass transition temperature, or a change in crystallinity. So far the partial effects of these influences cannot be separated. It is confirmed by experiment that in the last stages of drying, glass transition temperature of polymer changes rapidly. Also, the degree of crystallinity of the polymer increases during drying.3,4 From studies on polyaniline, it is known that its conductivity depends on the concentration of adsorbed molecules of water.5 Water interacts by hydrogen bonding with the polymer chain. The activation energy of hydrogen bonding is very low at 3-5 kcal/mol. Drying at 120oC reduced the amount of water molecules from 0.75 to 0.3 molecules per aniline unit. This change in water concentration drastically alters electrical conductivity which decreases by three orders of magnitude. Drying for two hours at 120oC did not result in complete removal of water. Given that the activation energy of hydrogen bonding is very low, the process of interaction is probably not the main barrier to removal of residual moisture. Also the relationship between conductivity and number of molecules of water is linear in the range from 0.15 to 0.75 molecules of water per aniline unit which means that there is no drastic change in the mechanism by which water participates in increasing the conductivity of polyaniline. Its conductivity simply depends on the distance between neighboring adsorbed molecules of water which apparently participate in the charge migration.

n-heptane molecules/styrene unit

1

16.1 Residual solvents in various products

1127

Figure 16.1.2. Blister formation in polyethylene containing originally 4000 ppm hexane. a - lateral surface 2.2 s after extrusion, b - cross-section after 28 s. [Adapted, by permission from R J Albalak, Z Tadmor, Y Talmon, AIChE J., 36, 1313 (1990).]

SEM studies contribute to an understanding of a major obstacle to residual solvent removal. Figure 16.1.2 shows two photographs of polyethylene strands extruded from a melt which initially contained 4000 ppm hexane. After a short period of time following the extrusion, blisters form which remain in the material and become enlarged until they break and release solvent. This blistering mechanism, determines the rate of residual solvent removal from the material. The rate of removal depends on bubble nucleation, temperature, and polymer rheological properties.6 Observing such mechanisms makes it easy to understand the principle involved in stripping solvents which became popular in recent inventions.7-9 Stripping solvents were used to improve the taste and odor properties and the oxidative thermal stability of thermoplastic ethylene polymers.7 Volatile components, such as products of degradation, solvent and monomer contribute to taste problems and odor formation. Striping solvents used include highly volatile hydrocarbons (ethylene, propylene, isobutane), inert gases, and supercritical fluids. An addition of at least 0.1% stripping solvent reduces volatiles from the typical levels of between 300-950 ppm to 45 ppm with even as low as 10 ppm possible. A stripping solvent helps in the generation of bubbles and their subsequent breaking by which both the stripping and the residual solvent are removed. In cosmetics and pharmaceutical formulations traces of solvents such as benzene or dichloromethane, used in the synthesis of acrylic acid polymer, disqualify the material. It is not unusual for this polymer to contain up to 1000 ppm of dichloromethane or up to 100 ppm of benzene. The use of mixed ester solvents helps to reduce residual solvent to below 5 ppm.8 Polycarbonate pellets from normal production may contain up to 500 ppm solvent. This makes processing polycarbonate to optical products very difficult because of bubble formation. Elimination of volatiles renders the product suitable for optical grade articles.9 These inventions not only demonstrate how to eliminate solvents but also confirm that the mechanism discussed in Figure 16.1.2 operates in industrial processes. The examples also show that large quantities of residual solvents are retained by products in their normal synthesis. Many standard methods are used to devolatilize materials. Flash devolatilizer or falling strand devolatilizer are synonyms of equipment in which the falling melt is kept below the saturation pressure of volatiles. Styrene-acrylonitrile copolymers devolatilized in flash devolatilizer had a final concentration of ethylbenzene of 0.04-0.06.1 Devolatilization of LLDPE in a single-screw extruder leaves 100 ppm of hydrocarbon solvent. 500 ppm chlorobenzene remains in similarly extruded polycarbonate.1 It is estimated that if the polymer contains initially 1-2% solvent, 50-70% of that solvent will be removed through the vacuum port of an extruder.1 These data seem to corroborate the information included in the above

1128

George Wypych

discussed patents. The data show that considerable amounts of residual solvents can MEK be found in polymers and plastic materials. In the food industry, residual solvents associated with packaging odors enter food products from two sources: packing materiFat-free als and printing inks.10 It is estimated that 14% fat P concentrations of residual solvents have recently decreased (from 2000 mg/ream in past to 1000 mg/ream).10 However, a new problem has become apparent in introducPA tion of low fat or no fat food.11 It was discovered11 that more customer complaints about odor were received for these 0 2000 4000 6000 8000 1 104 low fat baked goods products. Analysis shows that fat was a good solvent for Peak area volatiles (solvents) and consequently solFigure 16.1.3. Volatility of three printing solvents (PA propyl acetate, P - n-propanol, MEK - methyl ethyl vent odor was not detected because flavor ketone) from fat-free and fat-containing cookies. 20 µg perception is developed relative to the consolvent added to 2.5 g cookies. [Data from T Clark, Paper centration of gaseous flavor compounds. Film Foil Converter, 70, 11, 48 (1996).] Figure 16.1.3 shows the concentrations of three solvents as detected by gas chromatography. Substantially higher concentrations are detected in solvent-spiked fat-free cookies than in fat-containing cookies.11 In this experiment solvents were added to the cookies. In another experiment, packaged cookies were exposed to a solvent vapor atmosphere and different trends were recorded for cookies packaged in two different films. If the film had good barrier properties, no difference was noticeable between both types of cookies and the adsorbed quantities of solvents were minimal. If the cookies were packaged in a coextruded film having lower barrier properties, no-fat cookies absorbed 42% more solvent than full fat cookies. In the construction industry, residual solvent evaporation becomes an increasingly more critical issue, especially in the case of products used for indoor applications. Sealants, adhesives, and paints are now a major focus of this concern since they contribute to indoor pollution. Similar trends are observed in the automotive industry where both solvents and plasticizers are suspected of contributing to a “plastics” odor in car interiors. In contrast, changes in the solvent evaporation rate may also contribute to product improvement in paints. Small quantities of properly selected solvents can improve physical properties and the appearance of paints. Other applications of residual solvents include time-controlled release of fertilizers and production of materials with controlled morphology.

REFERENCES 1 2 3 4 5 6 7 8 9 10 11

R J Albalak, Ed., Polymer Devolatilization, Marcel Dekker, Inc., New York, 1996. L A Errede, P J Henrich, J N Schrolpfer, J. Appl. Polym. Sci., 54, 649 (1994). M O Ngui, S K Mallapragada, J. Polym. Sci., Polym. Phys., 36, 2771 (1998). H-T Kim, S-K Kim, J-K Park, Polym. J., 31, 154 (1999). E S Matveeva, Synthetic Metals, 79, 127 (1996). R J Albalak, Z Tadmor, Y Talmon, AIChE J., 36, 1313 (1990). M M Hughes, M E Rowland, C A Strait, US Patent 5,756,659, The Dow Chemical Company, 1998. A Bresciani, US Patent 5,342,911, 1994. T Hosomi, T Umemura, T Takata, Y Mori, US Patent 5,717,055, Mitsubishi Gas Chemical, Ltd., 1998. R M Podhajny, Paper Film Foil Converter, 72, 12, 24 (1998). T Clark, Paper Film Foil Converter, 70, 11, 48 (1996).

16.2 Residual solvents in pharmaceutical substances

1129

16.2 RESIDUAL SOLVENTS IN PHARMACEUTICAL SUBSTANCES Michel Bauer International Analytical Department, Sanofi-Synthélabo, Toulouse, France

Christine Barthélémy Laboratoire de Pharmacie Galénique et Biopharmacie, Faculté des Sciences Pharmaceutiques et Biologiques, Université de Lille 2, Lille, France

16.2.1 INTRODUCTION The need to test for residual solvents (RS) in pharmaceutical substances was recognized in the late 70’s when some pharmacopoeias like those of USA (USP XX) or Great-Britain (BP 80 + add 82) introduced specific tests for RS in some monographs. But we had to wait until the early 80’s to see a rational approach for establishing specifications from toxicological data. This strategy was developed by a working group of the Italian Pharmacopoeia1 starting from the threshold limit values for Chemical Substances and Physical Agents in the Work Environment published by the American Congress of Governmental Experts for Industrial hygiene.2 In the late 80’s, RS were definitively classified as impurities per se. Methods and specifications appeared in different issues of Pharmacopeial Forum and were submitted for discussion and finally integrated in the USP. At the same time only few such monographs could be found in the European Pharmacopoeia (Eur. Ph.) or British Pharmacopoeia (BP). Interestingly, the notion of a content limit for residual solvents in relation to the daily intake of the drug was introduced (see Eur. Ph. 2nd edition), a concept which was taken up later in the ICH Guideline. Numerous publications have been devoted to this subject.3-9 Although the number of papers on RS is immense, they are very often limited to specialized areas such as regulatory aspects or methodology aspects. In this chapter, different topics will be considered, starting with the fundamental question: why look for RS in pharmaceutical products? It is worth noting that Witschi and Doelker10 published in 1997 a very detailed and up-to-date review stressing the importance of this subject in the pharmaceutical field. 16.2.2 WHY SHOULD WE LOOK FOR RS? As we already have seen in Chapter 14.21.1, RS could have various effects on the drug substances, excipients and drug products. 16.2.2.1 Modifying the acceptability of the drug product The presence of RS could seriously impair customer compliance because of the odor or the taste they can cause in the final pharmaceutical preparation. Rabiant2 quotes the case of a drug substance having undergone, for technical reasons, a washing with isopropanol not planned in the manufacturing protocol. The oral solution prepared from this batch contained 100 ppm of this solvent and consequently had an odor that the majority of the patients accepted only with reluctance; the batch concerned was finally removed from the market.

1130


16.2.2.2 Modifying the physico-chemical properties of drug substances (DS) and drug products (DP) The role of the quality of the solvents on the stability e.g. of the raw materials, DS and DP (see Section 14.21.1) has already been discussed. It must be remembered that RS (including water) can show different kinds of interactions with solid substances:11 • Solvents adsorbed on the crystal surfaces, which generally are easily removed because of the existence of weak physical interactions. • Occluded solvents and clathrates which are more difficult to extract without impairing the quality of the drug by, for example, excessive drying. • Solvents bound to drug molecules in the crystal lattice. These solvents present as solvates (hydrates) are lost at a characteristic temperature and may be stable only over a limited range of relative humidity. The solvates and desolvated solvates, whilst being two different chemical entities, can retain the same crystalline structure (similarity of x-ray diffraction pattern) but show different physico-chemical properties.12-14 One of the most important effects of organic solvents absorbed on crystal surfaces is the ability to reduce the wettability of the crystals, especially if the solvent concerned is hydrophobic.15 Another interesting aspect, particularly in the case of residual water, is its role as an agent of recrystallization of poorly crystalline substances. It is well known that amorphous or partially amorphous products can undergo a recrystallization16,17 process over time in presence of water. If amorphous phases are an interesting way to promote the dissolution rate of poorly soluble drugs, their main drawback is their physical instability triggering the possible crystallization of the drug and leading to a decrease in dissolution rate and possibly of bioavailability, over a period of time. Furthermore the residual adsorbed water may have an impact on the flowability of a powder, which is linked to the solubility of the substance and the hydrophilicity of the crystal faces.15 Other physico-chemical parameters are influenced by RS, like particle size and dissolution properties. For more information we refer to the publications of Doelker10 and Guyot-Hermann15 and references quoted therein. Nevertheless, one example deserves to be mentioned here.10,16 Residual isopropyl alcohol enhanced the water permeability of Eudragit® L films used as tablet coating for protecting water-sensitive drugs as demonstrated by List and Laun.16 This implies particular conditions for the storage of coated tablets during the film-drying process. The atmosphere should be as dry as possible. The need to keep the RS level as low as possible can lead to some problems. It has been reported15,17 that a drug substance displaying a strong odor of residual solvent was submitted to reprocessing, consisting of the displacement of the residual solvent by a stream of water vapor. During this process which slightly modified the surface crystallinity of the particles, a small amount of an impurity was produced. The consequence was an increase of the surface solubility of this drug substance. During the preparation of the DP using an aqueous wet granulation, a liquification of the granulate was observed making the manufacture impossible. After having removed the impurity by purifying the DS a successful manufacture of the DP was achieved. There are other aspects linked to the manufacturing process and drying conditions which impact on the final RS content. They are discussed in Chapter 15.2.3 of this book (and references cited therein). Nevertheless before closing this paragraph, another example of the relationship between the manufacturing process and RS is worth mentioning here. It relates to the formation of volatile compounds produced by radiolysis and which could induce odor. Barbarin et


1131

al.18 have investigated this subject in different antibiotics belonging to the cephalosporin group (Cefotaxime, Cefuroxime and Ceftazidine). Using GC-MS and GC-IR they were able to identify carbon monoxide, nitric oxide, carbon disulfide, methanol, acetaldehyde, ethyl formate, methyl acetate and acetaldehyde O-methyloxime. In a subsequent publication19 on cefotaxime, they demonstrated that some of the radio-induced compounds (such as carbon monoxide sulfide (COS) and carbon disulfide) came from the degradation of the drug itself whereas the formation of others required the RS, present before the irradiation. For instance, acetaldehyde arises from the irradiation of methanol. Incidentally, this is a good way to differentiate between radio-sterilized and non-radio-sterilized products. It is worth remarking that residual humidity can favor15 microbiological growth especially in some natural products used as excipients (starch, gelatine, etc.). From a physico-chemical point of view, residual humidity may have an impact on the hardness of the tablets as shown by Chowhan,20 Down and McMullen.21 16.2.2.3 Implications of possible drug/container interactions It is possible that the RS contained in a powder may migrate up to the interface between the contents and the container facilitating the extraction and migration into the drug of additives used during the container manufacture. On the contrary, solvents may be used during the packaging of a drug. An example of this situation is described by Letavernier et al.22 Cyclohexanone was used for sealing PVC blisters containing suppositories. After 42 months of storage at ambient temperature up to 0.2 mg to 0.3 mg of cyclohexanone was found per gram of suppository. 16.2.2.4 As a tool for forensic applications Forensic laboratories are interested in identifying and assessing trace impurities in bulk pharmaceutical products with the idea of using the impurity profile as a “fingerprint” of the manufacturer. Of course, RS can be an important aid in this process.23,24 It has been demonstrated,25 for instance, that static headspace GS coupled with mass spectrometry (MS) was able to detect and identify volatile impurities, making possible the characterization of illicit heroin or cocaine samples. 16.2.2.5 As a source of toxicity 16.2.2.5.1 General points Toxicity is obviously the main reasons for testing for RS. Besides the toxicity of the drug itself, the related impurities,26 degradants and the RS obviously can each bring their own contribution. A drug could be prescribed to a patient either for a short period of time or a long one. To minimize either acute toxicity or chronic toxicity resulting from some accumulation process, RS have to be kept at the lowest achievable level. When developing new chemical entities, the presence of RS could bias toxicity studies including mutagenicity and carcinogenicity and cause a risk of wrongly ascribing to the drug substance or the formulation, side effects which are actually due to volatile impurities. Knowing the cost of such studies, it is preferable to have the RS under control. The long-term exposure to solvents has been recognized for a long time as a possible cause of serious adverse effects in human. Tables of maximal tolerated solvent concentrations in air for defined exposures have been published and used to set limits for residual organic solvents in pharmaceuticals.1-10 In the late 90’s, a set of articles appeared in the Pharmacopeial Forum proposing RS limits from toxicological data,27 including carcinogenicity, mutagenicity, teratogenicity and neurotoxicity.

1132


Table 16.2.1. (After from reference 28)

In the early 80’s, the “International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human use” (ICH) was created. Among the different topics deserving to be harmonized, the need to have an agreement amongst Europe, Japan and the USA on the ways to limit the RS in pharmaceuticals was clearly identified and the topic adopted in June 1994. The final guide was finally adopted in 1998 by the Health Authorities of the three zones and is now in force. This Guideline will be examined in detail in the last paragraph of this part. 16.2.2.5.2 Brief overview of the toxicology of solvents28,29 16.2.2.5.2.1 Aspects concerning metabolism There are four main routes by which solvents can interact with the human body: • O.R.L. (Otorhinolaryngology) • transdermal (including the ocular area) • oral • injection Once in the organism, solvents will undergo biotransformation which essentially takes place in the liver. This metabolism very often leads to more water-soluble products than the parent compound30,31 and, as such, more easily excreted by the kidneys. This detoxication process is beneficial for the individual but varies greatly from one subject to another. Unfortunately, this metabolic detoxication can be complicated by the appearance of reactive intermediates which, if not rapidly inactivated, will destroy the essential constituents of the cells (proteins, nucleic acids, unsaturated lipids) and cause INTOXICATION. The latter ranges from a simple allergic reaction to tissue necrosis or, at worst, to cancer. Table 16.2.1 summarizes the different events which can occur in the organism. 16.2.2.5.2.2 Solvent-related pathology 16.2.2.5.2.2.1 Acute toxicity It is especially the affinity of the solvents for lipid-rich organs which triggers problems of acute toxicity and these concern primarily the nervous system, the heart, the liver and kidneys. In this acute toxicological process, the molecules act per se without any previous biotransformation. The acute toxicity encompasses:


1133

• nervous toxicity (headache, somnolence, coma, more or less deep, which can extend to death)

• cardiac toxicity32 • action on skin and mucous membranes generating irritation (including ocular area)33 16.2.2.5.2.2.2 Long-term toxicity A prolonged exposure, even at low doses, to several liposoluble solvents leads sooner or later to irreversible effects on different organs: • central nervous system (e.g., toluene could lead to degeneration of the brain) • peripheral nervous system (e.g., methanol shows a peculiar affinity for the optic nerve leading possibly to blindness) • liver and kidneys; as solvents are metabolized in the liver and excreted by kidneys, these two organs are, of course, particular targets for these products • skin and mucous membranes (e.g., dermatitis) • blood (cyanosis, anaemia, chromosomal abnormality) With regard to carcinogenicity, benzene has long been recognized as carcinogen in man. It is the reason why its use is strictly limited and not recommended (ICH class I/specifications 2 ppm). Carbon tetrachloride and 1,2-dichloroethane have been demonstrated to be carcinogenic in animals and potentially carcinogenic in man. The embryotoxicity of solvents must be taken into account. As solvents can cross the placenta, pregnant women should be especially protected. 16.2.2.5.2.2.3 Metabolism of benzene By way of illustration, Figure 16.2.1 summarizes the metabolic pathway of benzene.34 Numerous publications dealing with the metabolism of benzene are given in reference 27.2. 16.2.3 HOW TO IDENTIFY AND CONTROL RS IN PHARMACEUTICAL SUBSTANCES? 16.2.3.1 Loss of weight Historically this is the first method which appeared in the pharmacopoeias, performed either at normal pressure or under vacuum. This is, of course, an easy method, particularly for routine control but there are several drawbacks: • lack of specificity • it is product demanding (1-2 g) • the limit of detection (LOD) is currently about 0.1 % This determination can now be done by thermogravimetric analysis (TGA) which makes the method more sensitive (possible LOD 100 ppm) and less product demanding (5 20 mg). It can also be used as a hyphenated method linking TGA to a mass spectrometer, allowing the identification of the desorbed solvents (specificity). However, whilst this kind of equipment exists, there are no signs of it replacing gas chromatography (GC) in the near future. 16.2.3.2 Miscellaneous methods Infrared spectroscopy35 and 1H-NMR36 have been used occasionally to identify and to quantify residual solvents, but their sensitivity is rather limited if compared with GC. On the other hand specificity is not always assured. The solvent should display signals well separated from those arising from the product, which is not always the case.

1134


Figure 16.2.1. Overview of benzene metabolism (after reference 34).

16.2.3.3 Gas chromatography (GC) 16.2.3.3.1 General points This is, of course, the method of choice which has long been used to determine RS whatever the area of application (pharmaceuticals, polymers, water analysis, etc.).37-40 From the late 70’s to the beginning of the 80’s, there was a large number (or flood) of publications dealing with different possible GC techniques which could be applied for detecting or analyzing residual solvents especially in pharmaceuticals (intermediates of synthesis, drug sub-


1135

stances, excipients and drug products). So much has been written that it is difficult to be original. Methodological aspects will be briefly covered in the next paragraph. The paper of Witschi and Doelker10 is particularly recommended. With 171 references, it represents a worthwhile, up-to-date review of the different GC techniques available. The last part of this chapter concerns pharmacopoeial methods. 16.2.3.3.2 Review of methods Regarding GC methodology, four aspects must be examined: • injection systems • columns • detectors • method validation 16.2.3.3.2.1 Injection systems 16.2.3.3.2.1.1 Direct injection After having dissolved the substance containing the RS to be looked for in an appropriate solvent, it is possible to directly inject into the system 100% of an aliquot of the solution (if packed columns are used) or partially through a split system (if capillary, narrow-bore and wide-bore columns are used). It is simple, accurate and repeatable (with an internal standard). The main drawback is that samples very often contain non-volatile substances which are retained by the column, leading rapidly to a loss of efficiency and a dramatic decrease in sensitivity.41 In the current literature dealing with the RS, the direct injection process is less frequently used. Nevertheless, publications have appeared until recently using split/ splitless injection.42,43 16.2.3.3.2.1.2 Static headspace injector The solubilized or suspended sample in an appropriate vehicle is heated at a defined temperature in a tightly closed vial until thermodynamic equilibrium is reached between the liquid phase and the gas phase. A known aliquot is then transferred either with the aid of a syringe or by an automatic transfer system onto the column. The main advantage is that only volatile products including solvents are injected into the column greatly improving its lifetime. The sensitivity is good and the system is easily automated. The main drawback is the existence of matrix effects and the possible non- ideality of the solvents mixture. These imply that ideally one should determine the calibration curve by adding standard solutions of the solvents of interest to the sample matrix, free of solvents. Because it is very difficult to obtain such a sample matrix, the classical standard addition method is recommended. It consists of adding to the sample matrix to be analyzed a known amount of the solvents to be determined. This method requires two analyses for the final calculation but the main advantage is that the matrix effect is overcome. The linearity of the response has, of course, to be demonstrated before the use of the simplified version mentioned above. Nevertheless, if based on a sound validation, external calibration can be used.41,44 The nature of the solvent used to prepare the solution (e.g., water, dimethylformamide, dimethylacetamide, 1,3-dimethyl-2-imidazolidinone (DMI)), equilibration temperature, the ratio between the gas phase and the liquid phase and the possible need to promote a salting out effect by adding mineral salts are the parameters, amongst others, which should be investigated and optimized to improve the sensitivity (LOD, LOQ) of the method.45-47 Another version of this static headspace chromatography is what has been called by Kolb48 multiple headspace extraction (MHE) chromatography. This is a multi-step injection

1136


technique which was alluded to in the Suzuki publication39 and more openly developed by MacAuliffe.49 The principle of this method is the following.50,51 After the first extraction has been made and the aliquot injected, the gas phase is removed by ventilating the vial and re-establishing the thermodynamic equilibrium. The equilibrium between the analyte in the solid or liquid phase and the gas phase will be displaced each time. After n extractions, the analyte content in the liquid or solid phase becomes negligible. It is then sufficient to sum the peak areas obtained for each extraction (which decrease exponentially) and, from an external calibration curve determine the amount of RS in the substance. This method is particularly useful for insoluble products or in cases where the partition coefficient of the RS is too favorable relative to the liquid phase. It has been recently successfully applied to the determination of RS in transdermal drug delivery systems.51 16.2.3.3.2.1.3 Dynamic headspace injection In the dynamic headspace method, the sample is put in a thermal desorption unit in order to desorb the RS; a continuous flow of a carrier gas pushes the RS into a trapping system which is refrigerated and where they are accumulated prior to analysis. Then the RS are rapidly desorbed by rapid heating and carried onto the column via the carrier gas. There are different ways to apply this technique.10,52 The arrangement when purge gas passes through the sample is often called the purge and trap technique (some other equipment uses the acronym DCI (desorption, concentration, injection)). This method is particularly useful for very low concentrations of RS as the total amount of a substance is extracted and can be applied directly to powders without need to dissolve them. The main drawback is that the dynamic headspace methods are not readily automated.41 16.2.3.3.2.1.4 Other techniques Several others techniques dealing with the injection problems have been developed. Among them the solid-phase microextraction method52-55 (SPME) and the full evaporation technique56 must be mentioned. According to Camarasu,53 the SPME technique seems to be very promising for RS determination in pharmaceuticals, with much better sensitivity than the static headspace technique. 16.2.3.3.2.2 Columns The wealth of publications dealing with RS determination by GC is so impressive that it is difficult to provide an exhaustive review. The interested reader will find plenty of information in the references quoted so far and in others recently published, mentioned below. Packed columns,57 wide or narrow bore columns, capillary columns, etc. have been used for RS determination. It is true to say that capillary columns and narrow bore columns are the most often mentioned techniques. Today it is almost certain that any user can find in the literature10 the stationary phase and the relevant conditions to resolve his RS problem, at least in terms of selectivity. By way of example, it has been shown by Brinkmann and Ebel58,59 how it was possible to screen 65 of the 69 solvents mentioned in the ICH Guideline (discussed in paragraph 16.2.4) using capillary columns filled with two stationary phases (DB 624 and Stabilwax) which basically constitute the strategy proposed by the European Pharmacopoeia. 16.2.3.3.2.3 Detectors The almost universally used detector is the flame ionization detector (FID) which works with all organic solvents but which is not selective. The mass spectrometer detector (MSD) is now more and more utilized.24,53,55,60,61 It can be either universal in its electron ionization (EI) mode or selective in its selective ion monitoring mode (SIM). Other detectors, selective


1137

and/or universal, can be used.10,62 Among them the electron capture detector has to be mentioned when looking for chlorinated solvents, even if its use is not straightforward. 16.2.3.3.2.4 Method validation Whatever the technique used for determining the RS content in pharmaceuticals, a thorough validation of the complete analytical process has to be conducted according to the ICH Guidelines [Text on “validation of analytical procedures” and “validation of analytical procedures: methodology”63]. For testing impurities in a quantitative manner the following items have to be completed: • specificity (or more appropriately selectivity) • accuracy • precision • repeatability • intermediate precision (first part of reproducibility) • limit of detection (LOD), limit of quantification (LOQ) • linearity of the response • the range which is the interval between the upper and lower concentration (amounts) of analyte in the sample (including these concentrations) for which it has been demonstrated that the analytical procedure has a suitable level of precision, accuracy and linearity. In the publications mentioned above,24,41-43,51-54,58,59,64 it is possible for reader to find experimental procedures to conduct validation efficiently. Attention is drawn to the fact that, when using the static headspace technique in particular some other parameters have to be investigated, such as:24,51 • ratio gas phase/solid or liquid phase • temperature of equilibrium between the two phases and time to reach it • temperature of the transfer line • pressurization time and sampling time when using a fully automated headspace injector Finally, some additional comments are worth making: • As regards reproducibility, it is true that the best way to assess it is to set up an inter-laboratory study. It is of course burdensome but this is probably the only way to infer reasonable suitability parameters for the routine quality control monograph (resolution, plate number, tailing factor, repeatability, LOD, LOQ) and specifications. It is worth noting that the latter should take the performance of the analytical method into account. But for those who cannot follow this approach, sound ruggedness testing has to be performed. Maris et al. have designed a method65 for evaluating the ruggedness of a gas chromatographic method for residual solvents in pharmaceutical substances. • When using the static headspace injector the possible matrix effect should be studied and can be evaluated by a statistical method.66,67 • One of the most important suitability parameters in case of RS determination is, of course, the LOQ (and LOD). In order to avoid an unrealistic value in the QC monograph, it is highly recommended to use a working limit of quantification WLOQ (and WLOD) which consists of determining a reasonable upper limit for LOQ (and LOD). In fact, the LOQ derived from the validation package is obtained in what we can call an ideal or optimized environment. A way for determining α (> 1) in the relationship:

1138


Figure 16.2.2. Head-space injection of the gaseous sample into the chromatographic system. Typical chromatogram of class 1 solvents using the conditions described for System A (European Pharmacopoeia method). Flame-ionization detector: calculation of H/h for 1,1,1-trichloroethane.

WLOQ = α LOQ is to calculate the LOQ according to the signal to noise ratio S/N method63,68 (see Figure 16.2.2) and to repeat this determination independently several times (e.g., n = 6) during the intermediate precision determination. If LOQm is the mean value and σ the standard deviation, a possible definition of WLOQ could be the upper limit of the one side confidence limit at a specified risk α (e.g., 0.05). WLOQ = LOQ m + t n −1, α σ where t is Student’s coefficient.


1139

16.2.3.3.3 Official GC methods for RS determination Current official GC methods are described in USP XXIII under chapter 467 “Organic volatile impurities”. Four methods (I, IV, V, VI) are mentioned. Methods I, V and VI are based on direct injection. They are suitable for water-soluble drugs and V for water insoluble drugs. Method IV describes the static headspace technique and is used for water soluble drugs. Method VI is very general and refers to the individual monograph which describes the chromatographic conditions ( injection, column, conditions) which should be used. The main characteristics of these four methods are summarized in Table 16.2.2. The European Pharmacopoeia69 used a two-tiered process based on two different columns: System A. Fused silica capillary or semi capillary column (30 m x 0.32 mm (ID) or 30 m x 0.53 mm (ID)) DB 264 (1.8 µm or 3 µm film thickness of the phase (6 per cent cyanopropylphenyl-94 percent dimethylpolysiloxane)) which is identical to the USP method V. System B. Fused silica capillary or semi-capillary column (30 m x 0.32 mm (ID) or 30 m x 0.53 mm (ID) DB-wax (0.25 µm film thickness of Macrogol 20000R)). Table 16.2.2. Gas chromatographic methods described in USP 23 (from ref. 10) Method

USP Method I: Direct GC injection

USP Method IV: Static HSC

USP Method V: Direct GC injection

Sample

Standardization

Column

Detector

Dissolved in water or another appropriate solvent

External

30 m x 0.53 mm ID, fused silica, with 5µm a crosslinked G27 stationary phase and A 5 m x 0.53 mm ID silica guard column, phenylmethyl siloxane deactivated

Dissolved in water containing sodium sulphate and heated for 1 h at 80°C before injection of the headspace

External

As USP Method V

FID

As in USP Method I

30 m x 0.53 mm ID, fused silica with 3 µm b G43 stationary phase and a 5 m x 0.5 mm ID silica guard column, phenylmethyl siloxane deactivated

FID

As in USP Method I

One of 9 columns , listed under , specified in the monograph

As USP Method I

FID*

c

USP benzene > chlorobenzene > p-xylene > nitrobenzene > o-dichlorobenzene > o-cresol > ethyl benzene. Methylene chloride and 1,1,1-trichlorethane would dominate the chlorinated hydrocarbons. Among the group of unrelated organic solvents, the concentration of pyridine at the well was predicted to increase rapidly. Pyridine would eventually dominate this group in the relative order: pyridine > carbon disulfide > ethyl acetate > diethyl ether. The relative order of fluorocarbons at the compliance well in terms of concentration was: F-21, F-22 >> F-12 > F-113 > F-114 > F-112 > R-112a > FC-115 >> F-11. In brief, the computer simulations predicted that all 28 organic compounds would eventually migrate from the waste trench, and be detected at the compliance well. The predicted concentrations varied by four orders of magnitude, and were largely influenced by the initial concentrations used in calculating the mass loading rate to the aquifer. 17.2.3 MASS LIMITATIONS The next step in this analysis was to determine whether these predicted concentrations would pose an environmental hazard by evaluating whether the site was capable of attenuating the concentrations of the organic compounds to levels that are protective of human health. In Figure 17.2.4 the predicted steady-state concentrations of the organic compounds

Figure 17.2.4. The predicted steady-state concentrations (Css) of each solvent in groundwater at the compliance point as a function of its Maximum Permissible (MPC) Concentration (Roy et al.1).

1168

William R. Roy

in groundwater at the compliance well were plotted against their MPCs. The boundary shown in Figure 17.2.4 represents the situation where the steady-state concentration (Css) equals the MPC. Consequently, the predicted Css is less than its corresponding MPC when the Css of a given compound plots in the lower-right side. In this situation, these organic compounds could enter the aquifer at a constant mass loading rate without exceeding the attenuation capacity of the site. The steady-state concentrations of twenty solvents exceeded their corresponding MPCs. The continuous addition of these organic compounds (i.e., a constant mass loading rate) would exceed the site's ability to attenuate them to environmentally acceptable levels in this worst-case scenario. There are two avenues for reducing the steady-state concentrations downgradient from the trench: (1) reduce the mass loading rate, and/or (2) reduce the mass of organic compound available to leach into the aquifer. Because, the RCRA-required double liner was regarded as the state-of-the-art with respect to liner systems, it was not technically feasible to reduce the volume of leachate seeping into the aquifer under the conditions imposed. The worst-case conditions could be relaxed by assuming a lower leachate head in the landfill or by providing a functional leachate-collection system. Either condition would be reasonable and would reduce the mass loading rate. Another alternative is to reduce the mass available for leaching. In the previous simulations, the mass available to enter the aquifer was assumed to be infinite. Solute transport models can be used to estimate threshold values for the amounts of wastes initially landfilled.2 A threshold mass (Mt) can be derived so that the down-gradient, steady-state concentrations will be less than the MPC of the specific compound, viz., Mt = V(MPC x 1000) t

[17.2.2]

where: Mt V MPC t

the threshold mass in g/hectare the volume of leachate entering the aquifer in L/yr/hectare the maximum permissible concentration as g/L, and time in years; the amount of time between liner breakthrough and when the predicted concentration of the compound in the compliance well equals its MPC.

Using this estimation technique, Roy et al.1 estimated mass limitations for the compounds that exceeded their MPCs in the simulations. They found that benzene, carbon tetrachloride, dichloromethane, pyridine, tetrachloroethylene, 1,1,1-trichloroethylene, trichloroethylene and all chlorinated fluorocarbons would require strict mass limitations (60% removal

OECD 301C, modified MITI test, adapted activated sludge

16

28 d = 58% removal

OECD 301C, modified MITI test, adapted activated sludge

16

1 d = 14% removal, 3 d = 19%, 5 d = 60%, 6 d = 100%

OECD 302B, modified Zahn-Wellens test, industrial non-adapted activated sludge

16

9 d = 100% removal

OECD 302B, modified Zahn-Wellens test, non-adapted activated sludge

16

14 d = 94% removal

OECD 301A, domestic secondary effluent sewage, DOC measured

16

1174

J Devillers, A Chezeau, A Cicolella, E Thybaud

Name [CAS RN], Values

Comments

BOD5 = 0.05 g/g (5.2% ThOD), BOD10 = 0.39 g/g (57% ThOD), BOD20 = 1.08 g/g (72% ThOD) BOD = 0.25 g/g, COD = 2.08 g/g

Ref. 16

Dutch standard method, adapted sewage

16

Diethylene glycol monobutyl ether acetate [124-17-4] BOD5 = 13.3% ThOD, BOD10 = 18.4%, BOD15 = 24.6%, BOD20 = 67%

16

Triethylene glycol monoethyl ether [112-50-5] 5 d = 8% bio-oxidation, 10 d = 47%, 15 d = 63%, 20 d = 71%

filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), fresh water

10

5 d = 1% bio-oxidation, 10 d = 10%, 15 d = 12%, 20 d = 22%

filtered domestic wastewater, non-acclimated seed, 3, 7, 10 mg/l (at least two), salt water

10

ThOD = 1.89 g/g, BOD5 = 0.05 g/g, %ThOD = 3, COD = 1.84 g/g, %ThOD = 97

effluent from a biological sanitary waste treatment plant, 20 ± 1°C, unadapted seed, BOD = APHA SM 219, COD = ASTM D 1252-67

11

COD removed = 96.6%

mixed culture, acclimation in a semi-continuous system

12

BOD5 = 1.15 mmol/mmol

acclimated mixed culture, estimated by linear regression technique from a 20-day test

17

Triethylene glycol monobutyl ether [143-22-6] COD removed = 96.3%

mixed culture, acclimation in a semi-continuous system

12

Propylene glycol monomethyl ether [107-98-2] 50% removal: 10000 mg/l

static, nominal concentrations, L = 33-75 mm

39

Menidia beryllina

96-h LC50 >10000 mg/l

static, sea water, nominal concentrations, L = 40-100 mm

39

semi-static, adult polyps

31

42

Fish

Ethylene glycol monoethyl ether acetate [111-15-9] Crustacea

Artemia salina

24-h TLm = 4000 mg/l

static, 24.5°C

10

Fish

Carassius auratus

24-h TLm = 160 mg/l

static, 20 ± 1°C, measured concentrations

34

Pseudomonas aeruginosa 4-m biocidal = 0-2%

tested in jet fuel and water mixtures

25

Sulfate-reducing bacteria 3-m biocidal = 0-2%


25

Candida sp.

4-m biocidal = 2-5%


25

Cladosporium resinae

4-m biocidal = 2-5%


25

Gliomastix sp.

4-m biocidal = 2-5%


25

Ethylene glycol monopropyl ether [2807-30-9] Bacteria Yeasts Fungi

Ethylene glycol monoisopropyl ether [109-59-1] Fungi


42-d NG = 10%

NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C

27

Carassius auratus

24-h TLm >5000 mg/l


34

Poecilia reticulata

7-d LC50 = 5466 mg/l

semi-static, 22 ± 1°C, rounded value

38

Fish

Ethylene glycol monobutyl ether [111-76-2]

Bacteria

Pseudomonas putida

16-h TGK = 700 mg/l

toxicity threshold, inhibition of cell multiplication

24

Pseudomonas aeruginosa

4-m biocidal = 1-2%


25

Sulfate-reducing bacteria

3-m biocidal = 1-2%


25

17.3 Environmental fate of glycol ethers

Species

Comments

Ref.

8-d TGK = 35 mg/l


26

Scenedesmus quadricauda

8-d TGK = 900 mg/l


26

Selenastrum capricornutum

7-d EC50 >1000 mg/l, NOEC = 125 mg/l, LOEC = 250 mg/l

growth rate inhibition

16

Candida sp.

4-m biocidal = 2-3%


25

4-m biocidal = 2-3% 42-d NG = 5%

tested in jet fuel and water mixtures NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C

25 27

Gliomastix sp.

4-m biocidal = 2-3%


25

Chilomonas paramaecium

48-h TGK = 911 mg/l


28

Uronema parduczi

20-h TGK = 463 mg/l


29

Entosiphon sulcatum

72-h TGK = 91 mg/l


30

Hydra vulgaris (syn. H. attenuata)

72-h LC50 = 690 mg/l

semi-static, adult polyps

31

Crassostrea virginica

24-h LC50 = 181 mg/l (143 - 228), 48-h LC50 = 160 mg/l (125 204), 72-h LC50 = 114 mg/l (93.9 - 138), 96-h LC50 = 89.4 mg/l (72 - 110)

static, 22 ± 1°C, 10 organisms/tank

43

static, nominal concentrations

32

Daphnia magna

24-h LC50 = 1720 mg/l, 24-h LC0 = 1140 mg/l, 24-h LC100 = 2500 mg/l 24-h EC50 = 1815 mg/l (1698-1940), 24-h EC0 = 1283 mg/l, 24-h EC100 = 2500 mg/l 48-h EC50 = 835 mg/l

static, nominal concentrations, immobilization

33

static, immobilization

16

24-h TLm = 1000 mg/l

static, 24.5°C

10

Blue-green Microcystis aeruginosa algae

Algae

Yeasts

Fungi

Protozoa

Coelenterates

Mollusca

Crustacea

1179


Artemia salina

Results

1180


Species

Crustacea

Results

Comments

Ref.

Panaeus setiferus

24-h LC50 > 430 mg/l, 48-h LC50 = 173 mg/l (123 - 242), 72-h LC50 = 147 mg/l static, average size = 1.5 cm (116 - 186), 96-h LC50 = 130 mg/l (104 - 162)

Carassius auratus

24-h TLm = 1700 mg/l

Leuciscus idus melanotus

48-h LC50 = 1575 mg/l, static (Juhnke) 48-h LC0 = 1350 mg/l 48-h LC100 = 1620 mg/l 48-h LC50 = 1395 mg/l, static (Lüdemann) 48-h LC0 = 1170 mg/l, 48-h LC100 = 1490 mg/l

35

Pimephales promelas

96-h LC50 = 2137 mg/l

static

16

Poecilia reticulata

7-d LC50 = 983 mg/l

semi-static, 22 ± 1°C, rounded value

38

Notropis atherinoides

72-h LC50 >500 mg/l

static

16

96-h LC50 = 1490 mg/l


39

Fish Lepomis macrochirus

Cyprinodon variegatus


24-h LC50 = 2950 mg/l, 96-h LC50 = 2950 mg/l 24-h LC50 (125 - 176), 48-h LC50 (107 - 147), 72-h LC50 (105 - 138), 96-h LC50 (100 - 133)

43

34

44

= 149 mg/l = 126 mg/l static, 22 ± 1°C

43


39


27

4-m biocidal = 10-17%


25

Sulfate-reducing bacteria 3-m biocidal = 10-17%


25

Candida sp.


25

Menidia beryllina

= 121 mg/l = 116 mg/l

96-h LC50 = 1250 mg/l

Ethylene glycol dibutyl ether [112-48-1] Fungi


42-d NG = 5%

Diethylene glycol monomethyl ether [111-77-3] Bacteria

Yeasts




Species

Algae

Fungi

Crustacea

Fish

1181

Results

Comments

Ref.

Scenedesmus subspicatus

72-h EC50 > 500 mg/l, 72-h EC20 > 500 mg/l, biomass 72-h EC90 > 500 mg/l

45

Selenastrum capricornutum

96-h EC50 > 1000 mg/l

biomass

45

4-m biocidal = 10-17% 42-d NG = 20%

tested in jet fuel and water mixtures NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C

25 27

Gliomastix sp.

4-m biocidal > 25%


25

Daphnia magna

24-h LC50 = 1495 mg/l (1300 - 1600), static, nominal concentrations 48-h LC50 = 1192 mg/l (1100 - 6500)

45

Oncorhynchus mykiss

96-h LC50 > 1000 mg/l

semi-static, nominal concentrations

45

Carassius auratus

24-h TLm >5000 mg/l


34

Pimephales promelas

24-h LC50 = 6400 (6200 - 6600), 48-h LC50 = 6000 (6000 -6100), 72-h LC50 = 6000 (6000 -6100), 96-h LC50 = 5700 (5600 - 5900)

static

45


39


27


Lepomis macrochirus

mg/l mg/l mg/l mg/l

96-h LC50 = 7500 mg/l

Diethylene glycol dimethyl ether [111-96-6] Fungi


42-d NG = 20%

Diethylene glycol monoethyl ether [111-90-0]

Bacteria

Vibrio fischeri

5-min EC50 = 1000 mg/l, reduction in light output, nominal 5-min EC50 = 1290 mg/l concentrations 15-min EC50 = 10954 mg/l (10592.8 - 11327.5) NG = no visible mycelial growth and spore germination, 1% glucose-mineral salts medium, 30°C

Fungi


42-d NG = 20%

Coelenterates

Hydra vulgaris (syn. H. attenuata)

72-h LC50 = 17000 mg/l semi-static, adult polyps

46

47

27

31

1182


Species

48

Orconectes immunis

96-h LC50 = 34700 mg/l flow-through, 15.5°C, 0.47 g, mea(29100 - 41400) sured concentrations

48

Artemia salina

24-h TLm >10000 mg/l

static, 24.5°C

10

Tanytarsus dissimilis

48-h LC50 = 18800 mg/l

static, 22.3°C, measured concentrations

48

Oncorhynchus mykiss

Amphibia

Ref.

Daphnia magna

EchinoderArbacia punctulata mata

Fish

Comments

48-h LC50 = 4670 mg/l static, 21.1°C, measured concentra(3620 - 6010), tions 48-h LC50 = 3340 mg/l static, 23.3°C, measured concentra(2120 - 5280) tions

Crustacea

Insecta

Results

4-h EC50 = 10661 mg/l static, marine, early embryo growth, (5576.5 - 20895.9) nominal concentrations 1-h EC50 = 3370 mg/l static, marine, sperm cell, nominal (3145.4 - 3610.8) concentrations 5-h EC50 = 4116 mg/l static, marine, increase in DNA, (3408 - 4907) nominal concentrations

47 47 49

96-h LC50 = 13400 mg/l flow-through, 14.9°C, 0.68 g, mea(11400 - 15700) sured concentrations

48

24-h TLm >5000 mg/l

34

Carassius auratus

static, 20 ± 1°C, measured concentrations 96-h LC50 = 20800 mg/l flow-through, 20.1°C, 0.92 g, mea(15700 - 27500) sured concentrations

Gambusia affinis

96-h LC50 = 15200 mg/l (12400 - 18700) 96-h LC50 = 12900 mg/l (11100 - 15000)

flow-through, 18.3°C, 0.23 g, measured concentrations flow-through, 19.9°C, 0.25 g, measured concentrations

48

Ictalurus punctatus

96-h LC50 = 6010 mg/l

flow-through, 17.3°C, 0.72 g, measured concentrations

48

Lepomis macrochirus

96-h LC50 >10000 mg/l static, nominal concentrations, L = 33-75 mm 96-h LC50 = 21400 mg/l flow-through, 18.3°C, 0.44 g, mea(19100 - 23900) sured concentrations

Pimephales promelas

96-h LC50 = 13900 mg/l (11600 - 16700) 96-h LC50 = 9650 mg/l (7910 - 11800)

flow-through, 24.8°C, 1.44 g, measured concentrations flow-through, 18.1°C, 0.35 g, measured concentrations

48

Menidia beryllina

96-h LC50 >10000 mg/l


39

Rana catesbeiana

96-h LC50 = 20900 mg/l flow-through, 17.2°C, 3.54 g, (19400 -22600) measured concentrations

48

39 48

48

Diethylene glycol diethyl ether [112-36-7] Fungi


42-d NG = 20%


27


Species

1183

Results

Comments

Ref.

toxicity threshold, inhibition of cell multiplication growth inhibition, 25 ± 2°C

24

Diethylene glycol monobutyl ether [112-34-5] 16-h TGK = 255 mg/l Bacteria

Pseudomonas putida 16-h EC10 = 1170 mg/l

Blue-green Microcystis aeruginosa algae

Algae

Fungi

Protozoa

Crustacea

50

8-d TGK = 53 mg/l


26

Scenedesmus quadricauda

8-d TGK = 1000 mg/l


26

Scenedesmus subspicatus

96-h EC50 > 100 mg/l

limit test

50


42-d NG = 5%


27

Chilomonas paramaecium

48-h TGK = 2774 mg/l


28

Uronema parduczi

20-h TGK = 420 mg/l


29

Entosiphon sulcatum

72-h TGK = 73 mg/l


30

static, nominal concentrations

32

static, nominal concentrations, immobilization

33

Daphnia magna

24-h LC50 = 2850 mg/l, 24-h LC0 = 1750 mg/l, 24-h LC100 = 3850 mg/l 24-h EC50 = 3200 mg/l (2990-3424), 24-h EC0 = 2333 mg/l, 24-h EC100 = 5000 mg/l 48-h EC50 >100 mg/l 48-h NOEC >100 mg/l

directive 84/449/ EEC, C2

16


34

Carassius auratus

24-h TLm = 2700 mg/l

35

Leuciscus idus melanotus

48-h LC50 = 1805 mg/l, static (Juhnke) 48-h LC0 = 1140 mg/l, 48-h LC100 = 2185 mg/l 48-h LC50 = 2304 mg/l, static (Lüdemann) 48-h LC0 = 1820 mg/l, 48-h LC100 = 2400 mg/l 48-h LC50 = 2750 mg/l static

Poecilia reticulata

7-d LC50 = 1149 mg/l

semi-static, 22±1°C, rounded value

38

Lepomis macrochirus

96-h LC50 = 1300 mg/l


39

Menidia beryllina

96-h LC50 = 2000 mg/l


39

Fish

35

16

1184


Species

Results

Comments

Ref.

Triethylene glycol monomethyl ether [112-35-6]

Bacteria

Yeasts Fungi




25




25

Candida sp.



25




25

Gliomastix sp.



25

Triethylene glycol monoethyl ether [112-50-5] Pseudomonas aeruginosa



25


3-m biocidal = 2-5%


25

Candida sp.



25




25

Gliomastix sp.



25

Crustacea

Artemia salina

24-h TLm >10000 mg/l

static, 24.5°C

10

Fish

Carassius auratus

24-h TLm >5000 mg/l


34


27

Bacteria

Yeasts Fungi

Propylene glycol monomethyl ether [107-98-2] Fungi


42-d NG = 20%

Propylene glycol monomethyl ether acetate [108-65-6] Fish

Lepomis macrochirus

24-h LC50 = 206 mg/l 96-h LC50 = 164 mg/l

44

It is difficult to draw definitive conclusions from the data listed in Table 17.3.2. Indeed, most of the data have been retrieved from rather old studies performed without GLP protocols. In addition, the toxicity values are generally based on nominal concentrations and the endpoints are different. However, from the data listed in Table 17.3.2, it appears that despite a difference of sensibility among species, glycol ethers do not present acute and subacute ecotoxicological effects to the majority of the tested organisms. However, it is interesting to note that the acetates seem to be more toxic that the corresponding parent compounds.10,34,39,51 In mammals, the acute toxicity of glycol ethers is also relatively low. The main target organs are the central nervous and haematopoitic systems. However, on the basis of the available data no significant difference exists between the acute toxicity of glycol ethers and their corresponding acetates.52


1185

17.3.4.2 Reproduction and development The reproductive and developmental toxicity of the ethylene glycol monomethyl and monoethyl ethers is well documented. Several longer-chain glycol ethers also have been investigated for their reproductive and developmental effects against rodents and rabbits.53-56 Conversely, there is a lack of information on the reproductive and developmental ecotoxicity of glycol ethers and their acetates. Bowden et al.31 have tested the teratogenic effects of four glycol ethers through their ability to inhibit the regeneration of isolated digestion regions of Hydra vulgaris (syn. H. attenuata). They have shown that the concentrations of ethylene glycol monomethyl, monoethyl, monobutyl, and diethylene glycol monoethyl ethers that were 50% inhibitory to regenerating digestive regions (IC50) after 72-h of exposure were 19,000, 1400, 540, and 19,000 mg/l, respectively. More specifically, at 10,000 mg/l of ethylene glycol monomethyl ether, the digestive regions regenerated the mouth and some tentacles. At 19,000 mg/l only tentacle buds were seen, while 38,000 mg/l produced disintegration of the coelenterates. Ethylene glycol monoethyl ether at 900 mg/l allowed the regeneration of the mouth, some tentacles and the basal disc. At 1900 mg/l four digestive regions showed wound healing while the remainder were dead. A concentration of 3700 mg/l was lethal to both polyps and digestive regions. At concentrations up to 370 mg/l of ethylene glycol monobutyl ether, digestive regions regenerated some tentacles and in some cases the basal disc. Normal wound healing only was observed at 740 mg/l while at 920 mg/l the wounds were healed but the region expanded. Last, the digestive regions at 10,000 mg/l of diethylene glycol monoethyl ether regenerated the mouth and some tentacles. At 20,000 mg/l only tentacle buds were seen while a concentration of 40,000 mg/l was lethal to both polyps and digestive regions. Using the LC50 (Table 17.3.2)/IC50 ratio as developmental hazard index, Bowden et al.31 ranked the four studied glycol ethers as follows: Ethylene glycol ethyl ether (1.7) > ethylene glycol monomethyl ether (1.5) > ethylene glycol monobutyl ether (1.3) > diethylene glycol monoethyl ether (0.9). Johnson et al.57,58 have also ranked glycol ethers according to the difference between their lowest concentrations overtly toxic to adults (A) and their lowest concentrations interfering with development (D) of the artificial embryos of reaggregated adult Hydra attenuata cells. The A/D ratios found by these authors were the following: Ethylene glycol monoethyl ether (5.0) > ethylene glycol monobutyl ether (4.4) > ethylene glycol monophenyl ether (3.3) > diethylene glycol dibutyl ether (2.3) > diethylene glycol monoethyl ether (2.2) > ethylene glycol monomethyl ether (1.3) > ethylene glycol monomethyl ether acetate (1.0) = ethylene glycol monoethyl ether acetate (1.0). Daston et al.37 have shown that A/D ratios were not constant across species and hence, there was no basis for using this parameter for developmental hazard assessment. Thus, for example, the A/D ratios calculated from the lowest observed effect levels (LOELs) of the ethylene glycol monomethyl ether were 8, >3, 0.5, and 4, ≥6, 0.4, and ≤0.3 for the mammal, amphibian, fish, and insect, respectively. Teratogenicity of glycol ethers has been deeply investigated on the fruit fly, Drosophila melanogaster. Statistically significant increases in the incidence of wing notches and bent humeral bristles have been observed in Drosophila melanogaster exposed during development to ethylene glycol monomethyl ether (12.5, 15, 18, 22, and 25 mg/vial) and ethylene glycol monoethyl ether (54, 59, 65, 71, and 78 mg/vial).59 Wing notches, rare in control flies, were found in 13.8% of flies treated with ethylene glycol monomethyl ether

1186


(7.5 µl/g).60 In general, male pupae are much more affected by ethylene glycol monomethyl ether than female pupae. However, teratogenicity appears strain dependent. Higher detoxification occurs with increased alcohol dehydrogenase (ADH) activity. Ethylene glycol monomethyl ether is much more toxic than its oxidation product, methoxyacetic acid, at the level of adult eclosion. Teratogenic effects were observed in an ADH-negative strain in spite of lacking ADH activity suggesting that apparently, ethylene glycol monomethyl ether is a teratogenic compound by itself against Drosophila melanogaster.61 Last, it is interesting to note that recently, Eisses has shown62 that administration of ethylene glycol monomethyl ether to larvae of fruit fly, containing the highly active alcohol dehydrogenase variant ADH-71k, exposed the mitotic germ cells and the mitotic somatic cells of the imaginal discs simultaneously to the mutagen methoxyacetaldehyde and the teratogen methoxyacetic acid, respectively. Consequently, the chances for specific gene mutations, though non-adaptive, were likely increased by a feedback mechanism. 17.3.5 CONCLUSION Despite their widespread use, glycol ethers and their acetates have received little attention as potential environmental contaminants. Based on their physico-chemical properties, they would tend to remain in the aquatic ecosystems where their bioconcentration, biomagnification and sorption onto sediments will appear negligible. Volatilization from water and hydrolysis or photolysis in the aquatic ecosystems are generally of minimal importance. Glycol ethers are also poorly sorbed to soil and their rapid removal in the atmosphere is expected. While glycol ethers are biodegradable under aerobic conditions, these chemicals may persist for many years under anaerobic conditions. Based on the available acute ecotoxicity data, glycol ethers and their acetates can be considered as practically non-toxic. However, there is a lack of information on their long-term effects on the biota. This is particularly annoying because the developmental toxicity of some of them has been clearly identified against mammals. Consequently, there is a need for studies dealing with the potential long-term effects of these chemicals against organisms occupying different trophic levels in the environment in order to see whether or not the classical methodological frameworks used for assessing the environmental risk of xenobiotics remain acceptable for this class of chemicals. 17.3.6 ACKNOWLEDGMENT This study was supported by the French Ministry of the Environment as part of the PNETOX program (1998). REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13

R.J. Smialowicz, Occup. Hyg., 2, 269 (1996). Anonymous in Ethers de Glycols. Quels Risques pour la Santé?, INSERM, Paris, 1999, pp. 1-19. K. Nagano, E. Nakayama, M. Koyano, H. Oobayashi, H. Adachi, and T. Yamada, Jap. J. Ind. Health, 21, 29 (1979). Anonymous in Ethers de Glycols. Quels Risques pour la Santé?, INSERM, Paris, 1999, pp. 111-137. A. Cicolella, Cahiers de Notes Documentaires, 148, 359 (1992). W. Eckel, G. Foster, and B. Ross, Occup. Hyg., 2, 97 (1996). D.K. Nguyen, A. Bruchet, and P. Arpino, J. High Resol. Chrom., 17, 153 (1994). A. Yasuhara, H. Shiraishi, M. Tsuji, and T. Okuno, Environ. Sci. Technol., 15, 570 (1981). P. Ciccioli, E. Brancaleoni, A. Cecinato, R. Sparapani, and M. Frattoni, J. Chromatogr., 643, 55 (1993). K.S. Price, G.T. Waggy, and R.A. Conway, J.Water Pollut. Control Fed., 46, 63 (1974). A.L. Bridié, C.J.M. Wolff, and M. Winter, Water Res., 13, 627 (1979). T. Fuka, V. Sykora, and P. Pitter, Sci. Pap. Inst. Chem. Technol. Praze Technol. Water, F25, 203 (1983) (in Czech). S. Takemoto, Y. Kuge, and M. Nakamoto, Suishitsu Odaku Kenkyu, 4, 22 (1981) (in Japanese).


14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46

47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62

1187

P. Pitter and J. Chudoba, Biodegradability of Organic Substances in the Aquatic Environment, CRC Press, Boca Raton, 1990. G.T. Waggy, R.A. Conway, J.L. Hansen, and R.L. Blessing, Environ. Toxicol. Chem., 13, 1277 (1994). C.A. Staples, R.J. Boatman, and M.L. Cano, Chemosphere, 36, 1585 (1998). L. Babeu and D.D. Vaishnav, J. Ind. Microbiol., 2, 107 (1987). S.J. Gonsior and R.J. West, Environ. Toxicol. Chem., 14, 1273 (1995). T. Harada and Y. Nagashima, J. Ferment. Technol., 53, 218 (1975). F. Kawai, Appl. Microbiol. Biotechnol., 44, 532 (1995). R. Atkinson, Int. J. Chem. Kinetics, 19, 799 (1987). P.H. Howard, R.S. Boethling, W.F. Jarvis, W.M. Meylan, and E.M. Michalenko, Handbook of Environmental Degradation Rates, CRC Press, Boca Raton, 1991. D. Grosjean, J. Air Waste Manage. Assoc., 40, 1397 (1990). G. Bringmann and R. Kühn, Z. Wasser Abwasser Forsch., 10, 87 (1977). R.A. Neihof and C.A. Bailey, Appl. Environ. Microbiol., 35, 698 (1978). G. Bringmann and R. Kühn, Mitt. Internat. Verein. Limnol., 21, 275 (1978). K.H. Lee and H.A. Wong, Appl. Environ. Microbiol., 38, 24 (1979). G. Bringmann, R. Kühn, and A. Winter, Z. Wasser Abwasser Forsch., 13, 170 (1980). G. Bringmann and R. Kühn, Z. Wasser Abwasser Forsch., 13, 26 (1980). G. Bringmann, Z. Wasser Abwasser Forsch., 11, 210 (1978). H.C. Bowden, O.K. Wilby, C.A. Botham, P.J. Adam, and F.W. Ross, Toxic. in Vitro, 9, 773 (1995). G. Bringmann and R. Kühn, Z. Wasser Abwasser Forsch., 10, 161 (1977). G. Bringmann and R. Kühn, Z. Wasser Abwasser Forsch., 15, 1 (1982). A.L. Bridié, C.J.M. Wolff, and M. Winter, Water Res., 13, 623 (1979). I. Juhnke and D. Lüdemann, Z. Wasser Abwasser Forsch., 11, 161 (1978). S. Tsuji, Y. Tonogai, Y. Ito, and S. Kanoh, Eisei Kagaku, 32, 46 (1986) (in Japanese). G.P. Daston, J.M. Rogers, D.J. Versteeg, T.D. Sabourin, D. Baines, and S.S. Marsh, Fund. Appl. Toxicol., 17, 696 (1991). H. Könemann, Toxicology, 19, 209 (1981). G.W. Dawson, A.L. Jennings, D. Drozdowski, and E. Rider, J. Hazard. Materials, 1, 303 (1975-1977). K.L.E. Kaiser and V.S. Palabrica, Water Poll. Res. J. Canada, 26, 361 (1991). IRCHA, Les Produits Chimiques dans l’Environnement. Registre des Données Normalisées de leurs Effets dans l’Environnement. Classeurs I et II et mises à jour, IRCHA, Vert-Le-Petit, 1981-1985. J. Hermens, H. Canton, P. Janssen, and R. de Jong, Aquat. Toxicol., 5, 143 (1984). Results from the MBA Laboratories, Houston, Texas (1984). W.B. Neely, Chemosphere, 13, 813 (1984). IUCLID Data Set, RIVM/ACT, Substance ID: 111-77-3, 19-Nov-98. C. Curtis, A. Lima, S.J. Lozano, and G.D. Veith, in Aquatic Toxicology and Hazard Assessment. Fifth Conference. ASTM STP 766, J.G. Pearson, R.B. Foster, and W.E. Bishop, Eds., American Society for Testing and Materials, Philadelphia, pp. 170-178. D. Nacci, E. Jackim, and R. Walsh, Environ. Toxicol. Chem., 5, 521 (1986). R.V. Thurston, T.A. Gilfoil, E.L. Meyn, R.K. Zajdel, T.I. Aoki, and G.D. Veith, Water Res., 19, 1145 (1985). E. Jackim and D. Nacci, Environ. Toxicol. Chem. 5, 561 (1986). IUCLID Data Set, RIVM/ACT, Substance ID: 112-34-5, 19-Nov-98. R.B. Sleet, Toxicologist, 11, 296 (1991). Anonymous in Ethers de Glycols. Quels Risques pour la Santé, INSERM, Paris, 1999, pp. 51-67. C.A. Kimmel, Occup. Hyg., 2, 131 (1996). A. Cicolella, Santé Publique, 2, 157 (1997). Anonymous in Ethers de Glycols. Quels Risques pour la Santé?, INSERM, Paris, 1999, pp. 111-137. Anonymous in Ethers de Glycols. Quels Risques pour la Santé?, INSERM, Paris, 1999, pp. 139-162. E.M. Johnson, B.E.G. Gabel, and J. Larson, Environ. Health Perspect., 57, 135 (1984). E.M. Johnson, L.M. Newman, B.E.G. Gabel, T.F. Boerner, and L.A. Dansky, J. Am. Coll. Toxicol., 7, 111 (1988). D. Lynch and M. Toraason, Occup. Hyg., 2, 171 (1996). R.L. Schuler, B.D. Hardin, and R.W. Niemeier, Teratogenesis Carcinog. Mutagen., 2, 293 (1982). K.T. Eisses, Teratogenesis Carcinog. Mutagen., 9, 315 (1989). K.T. Eisses, Teratogenesis Carcinog. Mutagen., 19, 183 (1999).

1188

Michelle Bergin, Armistead Russell

17.4 ORGANIC SOLVENT IMPACTS ON TROPOSPHERIC AIR POLLUTION Michelle Bergin and Armistead Russell Georgia Institute of Technology, Atlanta, Georgia, USA

17.4.1 SOURCES AND IMPACTS OF VOLATILE SOLVENTS Solvents, either by design or default, are often emitted in to the air, and the total mass of emissions of solvents is not small. In a typical city in the United States, solvents can rival automobile exhaust as the largest source category of volatilized organic compound (VOC) emissions into the atmosphere.1 In the United Kingdom, solvent usage accounted for 36% of the estimated total VOC mass emissions in 1995.2 Such widespread emissions leads to increased concentrations of many different compounds in the ambient environment, and their release has diverse impacts on air quality. A large variety of solvent-associated compounds are emitted, many of which are hydrocarbons, oxygenates. Those solvents may have multiple atmospheric impacts. For example, toluene is potentially toxic and can reach relatively high concentrations at small spatial scales, such as in a workplace. Toluene also contributes to the formation of tropospheric ozone at urban scales, while at regional scales toluene can lower the rate of tropospheric ozone formation. Other solvents likewise can have a range of impacts, ranging from local contamination to modification of the global climate system. This diversity of potential impacts is due, in part, to differences in the chemical properties and reactions that a compound may undergo in the atmosphere, differences in emissions patterns, and differences in the spatial and temporal scales of atmospheric phenomena. Transport and fate of chemical species is closely tied to the speed at which the compound degrades (from seconds to centuries, depending on the compound) as well as to the environmental conditions in which the compound is emitted. If a compound degrades very quickly, it may still have toxic effects near a source where concentrations can be high. In contrast, extremely stable compounds (such as chlorofluorocarbons; CFCs) are able to circumvent the globe, gradually accumulating to non-negligible concentrations.3 Of the myriad of solvents emitted into the air, the ones of primary concern are those with the greatest emissions rates, and/or those to which the environment has a high sensitivity. Compounds with very large emissions rates include tri- and tetrachloroethylene (e.g., from dry-cleaning), aromatics (benzene, toluene and xylenes, e.g., from coatings), alcohols, acetone and, historically, CFCs. While those compounds are often emitted from solvent use, other applications lead to their emission as well. For example, gasoline is rich in aromatics and alkanes, and in many cases fuel use dominates emissions of those compounds. CFCs have been used as refrigerants and as blowing agents. This diversity of originating sources makes identifying the relative contribution of solvents to air quality somewhat difficult since there are large uncertainties in our ability to quantify emissions rates from various source categories. Solvents with a high environmental sensitivity include benzene (a potent carcinogen), xylenes (which are very effective at producing ozone), formaldehyde (both toxic and a strong ozone precursor), and CFCs (ozone depleters and potential greenhouse gases). Most of the solvents of concern in terms of impacting ambient air are organic, either hydrocarbons, oxygenated organics (e.g., ethers, alcohols and ketones) or halogenated organics (e.g., dichlorobenzene). Some roles of these compounds in the atmosphere are discussed below.

17.4 Organic solvent impacts on tropospheric air pollution

1189

While the toxicity of some solvents is uncertain, the role of emissions on direct exposure is not in question. Indoors, vaporized solvents can accumulate to levels of concern for acute and/or chronic exposure. However, the toxicity of solvents outdoors is not typically of as great of concern as indoors except very near sources. Outdoors, solvents have adverse effects other than toxicity. The importance of CFC emissions on stratospheric ozone, for example, is significant, but the problem is well understood and measures are in place to alleviate the problem. Reactive compounds can also aid in the formation of other pollutants, referred to as secondary pollutants because they are not emitted, but formed from directly emitted primary precursors. Of particular concern is tropospheric ozone, a primary constituent of photochemical smog. In the remainder of this chapter, the impacts of solvents on air quality are discussed, with particular attention given to the formation of tropospheric ozone. This emphasis is motivated by current regulatory importance as well as by lingering scientific issues regarding the role of volatile organics in secondary pollution formation. 17.4.2 MODES AND SCALES OF IMPACT Many organic solvents are toxic, and direct exposure to the compound through the atmosphere (e.g., via inhalation) can be harmful. While toxic effects of solvents rely on direct exposure, many solvents also contribute to the formation of secondary pollutants such as tropospheric ozone or particulate matter (PM), which cause health problems and damage the environment on larger spatial scales such as over urban areas and multi-state/country regions. Very slowly reacting solvent compounds also impact the atmosphere on the global scale, which may cause imbalances in living systems and in the environment. While some mechanisms of environmental imbalance are understood, the risks associated with global atmospheric impacts are highly uncertain. Transport of solvents in the atmosphere is similar to most other gaseous pollutants, and is dominated by the wind and turbulent diffusion. There is little difference between the transport of different solvent compounds, and the fact that most solvents have much higher molecular weights than air does not lead to enhanced levels at the ground. Heavy solvents are, for the most part, as readily diffused as lighter solvents, although they may not vaporize as fast. The higher levels of many solvents measured near the ground are due to proximity to emissions sources, which are near the surface, and the fact that most solvents degrade chemically as they mix upwards. A major difference in the evolution of various solvents is how fast they react chemically. Some, such as formaldehyde, have very short lifetimes while others, such as CFCs, last decades. 17.4.2.1 Direct exposure Volatilization of solvents allows air to serve as a mode of direct exposure to many compounds known to be toxic. Generally, direct exposure is a risk near strong or contained sources, and can cause both acute and chronic responses. Most of the non-workplace exposure to solvents occurs indoors. This is not surprising since, on average, people spend a vast majority of their time indoors, and solvents are often used indoors. Outdoors, solvents rapidly disperse and can oxidize, leading to markedly lower levels than what is found indoors near a source. For example, indoor formaldehyde levels are often orders of magnitude greater than outdoors. There still are cases when outdoor exposure may be non-negligible, such as if one spends a significant amount of time near a major source. Toxic effects of solvents are fairly well understood, and many countries have developed regulatory structures to protect people from direct exposure. The toxic effects of solvent emissions on ecosystems are less well understood, but are of growing concern.

1190


17.4.2.2 Formation of secondary compounds In addition to transport, organic compounds emitted into the air may also participate in complex sets of chemical reactions. While many of these reactions “cleanse” the atmosphere (most organic compounds ultimately react to form carbon dioxide), a number of undesirable side effects may also occur. Such adverse impacts include the formation of respiratory irritants and the destruction of protective components of the atmosphere. Ozone is a classic example of the complexity of secondary atmospheric impacts. Ozone is a highly reactive molecule consisting of three oxygen atoms (O3). In one part of the atmosphere ozone is beneficial, in another, it is a pollutant of major concern. Solvents and other organic emissions may either increase or decrease ozone concentrations, depending on the compound, location of reaction, and background chemistry. The mechanisms of some adverse secondary responses are discussed below. 17.4.2.3 Spatial scales of secondary effects Two layers of the Earth’s atmosphere are known to be adversely impacted by solvents - the troposphere and the stratosphere. These two layers are closest to Earth, and have distinct chemical and physical properties. The troposphere (our breathable atmosphere) is the closest layer, extending from the Earth to a height of between 10 to 15 km. The rate of chemical reaction generally determines the spatial scale over which emissions have an impact in the troposphere. Most non-halogenated solvents have lifetimes of a week or less, and elevated concentrations will only be found near the sources.4 Compounds that do not react rapidly in the troposphere (e.g., CFCs) are relatively uniformly distributed, and may eventually reach the stratosphere. The stratosphere is the next vertical layer of the atmosphere, extending from the tropopause (the top of the troposphere) to about 50 km in altitude. Little vertical mixing occurs in the stratosphere, and mixing between the troposphere and the stratosphere is slow. Impacts on the stratosphere can be considered global in scale, while impacts on the troposphere are generally urban or regional in scale. Distinct chemical systems of interest concerning solvents in the atmosphere are stratospheric ozone depletion, global climate change, and tropospheric photochemistry leading to enhanced production of ozone, particulate matter, and other secondary pollutants such as organonitrates. 17.4.2.3.1 Global impacts Because some solvent compounds are nearly inert, they can eventually reach the stratosphere where they participate in global scale atmospheric dynamics such as the destruction of stratospheric ozone and unnatural forcing of the climate system. Stratospheric ozone depletion by chlorofluorocarbons (CFCs) is a well-known example of global scale impacts. CFCs were initially viewed as environmentally superior to organic solvents. They are generally less toxic than other similarly acting compounds, less flammable and are virtually inert in the troposphere. Replacing solvents using volatile organic compounds (VOCs) with CFCs was hoped to reduce the formation of tropospheric ozone and other secondary pollutants. Because of their inert properties, there are no effective routes for the troposphere to remove CFCs, and, over the decades, emissions of CFCs have caused their accumulation, enabling them to slowly leak into the stratosphere. In the stratosphere, the strong ultraviolet (UV) light photodissociates CFCs, releasing chlorine, which then catalytically attacks ozone. CFC use has been largely eliminated for that reason. Partially halogenated organic solvents do not contribute as seriously to this problem since they react faster in the tropo-


1191

sphere than CFCs, so the associated chlorine does not reach the stratosphere as efficiently. CFCs and other solvent compounds also have a potential impact on global climate change. 17.4.2.3.2 Stratospheric ozone depletion Natural concentrations of stratospheric O3 are balanced by the production of ozone via photolysis of oxygen by strong UV light, and destruction by a number of pathways, including reactions with nitrogen oxides and oxidized hydrogen products that are present. Photolysis of an oxygen molecule leads to the production of two free oxygen atoms: O2 + hν —> O + O

[17.4.1]

Each oxygen atom can then combine with an oxygen molecule to form ozone: O + O2 —> O3

[17.4.2]

Ozone is then destroyed when it reacts with some other compound, e.g., with NO: O3 + NO —> NO2 + O2

[17.4.3]

Addition of either chlorine or bromine atoms leads to extra, and very efficient, pathways for ozone destruction. The free chlorine (or bromine) atom reacts with ozone, and the product of that reaction removes a free oxygen atom:

Net (reactions 4+5 together):

Cl + O3 —> ClO + O2

[17.4.4]

ClO + O —> Cl + O2

[17.4.5]

O3+ O —> 2 O2

[17.4.6]

Removing a free oxygen atoms also reduce ozone since one less ozone molecule will be formed via reaction 17.4.2. Thus, the chlorine atom reactions effectively remove two ozone molecules by destroying one and preventing the formation of another. Additionally, the original chlorine atom is regenerated to catalytically destroy more ozone. This reaction cycle can proceed thousands of times, destroying up to 100,000 molecules of O3 before the chlorine is removed from the system (e.g., by the formation of HCl). Reduction of ozone is greatly enhanced over the poles by a combination of extremely low temperatures, decreased transport and mixing, and the presence of polar stratospheric clouds that provide heterogeneous chemical pathways for the regeneration of atomic chlorine. The resulting rate of O3 destruction is much greater than the rate at which it can be naturally replenished. Current elevated levels of CFCs in the troposphere will provide a source of chlorine to the stratosphere for decades, such that the recent actions taken to reduce CFC emissions (through the Montreal Protocol) will have a delayed impact. 17.4.2.3 Global climate forcing Over the past decade, the potential for non-negligible changes in climate caused by human activity has been an issue of great concern. Very large uncertainties are associated with both estimations of possible effects on climate as well as estimations of the potential impacts of changes in climate. However, current consensus in the international scientific community is that observations suggest “a discernible human influence on global climate”.3

1192


Solvent compounds, especially CFCs and their replacements, participate in climate change as “greenhouse gases”. Greenhouse gases allow short-wave solar radiation to pass through, much of which the earth absorbs and re-radiates as long-wave radiation. Greenhouse gases absorb the long-wave radiation, causing the atmosphere to heat up, thereby acting as a blanket to trap radiation that would normally vent back to space. Climate change is a controversial and complex issue, but it is likely that restrictions such as those from the Kyoto Protocol will be adopted for emissions of compounds strongly suspected of exacerbating climate change. Many countries have already adopted stringent policies to reduce greenhouse gas emissions. 17.4.2.4 Urban and regional scales Another area of concern regarding outdoor air is exposure to secondary pollutants that are due, in part, to chemical reactions involving solvent compounds. Examples include the formation of elevated levels of ozone, formaldehyde, organonitrates, and particulate matter. Formaldehyde, a suspected carcinogen, is an oxidation product of organic compounds. Tropospheric ozone and organonitrates, as discussed below, are formed from a series of reactions of organic gases and nitrogen oxides in the presence of sunlight. Particulate matter formation is linked to ozone, and some solvents may react to form particulate matter. The particulate matter of concern is small (generally less than 2.5 µm in diameter) usually formed by gas-to-aerosol condensation of compounds via atmospheric chemical reactions. Ozone and particulate matter are both regulated as “criteria” pollutants in the United States because they have been identified as risks to human health. Ozone is believed to cause respiratory problems and trigger asthma attacks, and PM has a variety of suspected adverse health outcomes (e.g., respiratory and coronary stress and failure). Many organonitrates, such as peroxyacetyl nitrate, are eye irritants and phytotoxins. Currently, the formation and effects of ozone are better understood than those of fine particulate matter and organonitrates. The following section of this chapter discusses the effects, formation, and control of tropospheric ozone. The role of solvents in forming particulate matter is currently viewed as less urgent. 17.4.3 TROPOSPHERIC OZONE Tropospheric ozone, a primary constituent of photochemical smog, is naturally present at concentrations on the order of 20-40 parts per billion (ppb).4 However, elevated levels of ground-level ozone are now found virtually worldwide, reaching in some cities concentrations of up to 10 times the natural background. 17.4.3.1 Effects Ozone is believed to be responsible for both acute (short-term) and chronic (long-term) impacts on human health, especially on lung functions. Major acute effects of ozone are decreased lung function and increased susceptibility to respiratory problems such as asthma attacks and pulmonary infection. Short-term exposure can also cause eye irritation, coughing, and breathing discomfort.5-7 Evidence of acute effects of ozone is believed to be “clear and compelling”.8 Chronic health effects may present a potentially far more serious problem; however, definitive evidence is difficult to obtain. Recent studies do suggest that ambient levels of ozone induce inflammation in human lungs, which is generally accepted as a precursor to irreversible lung damage,6 and chronic animal exposure studies at concentrations within current ambient peak levels indicate progressive and persistent lung function and structural abnormalities.5,8


1193

Crop damage caused by air pollution has also received much attention. It is estimated that 10% to 35% of the world’s grain production occurs in regions where ozone pollution likely reduces crop yields.9 Air pollution accounts for an estimated several billion dollar crop loss every year in the United States alone, and research and analysis suggests that about 90% of this crop loss can be directly or indirectly attributed to ozone.10 Evidence also indicates that ozone may cause short- and long-term damage to the growth of forest trees,11 as well as altering the biogenic hydrocarbon emissions of vegetation.12 17.4.3.2 Tropospheric photochemistry and ozone formation In the lowest part of the atmosphere, chemical interactions are very complex. A large number of chemical compounds are present, the levels of many of these compounds are greatly elevated, and emissions vary rapidly due to both natural and anthropogenic sources. Ozone formation in the troposphere results from non-linear interactions between NOx, VOCs, and sunlight.4,13 In remote regions, ozone formation is driven essentially by methane,14 however elsewhere most VOCs participate in ozone generation. For example, measurements of non-methane organic compounds in the South Coast Air Basin of California during the 1987 Southern California Air Quality Study, identified more than 280 ambient hydrocarbon and oxygenated organic species,15 many of which originated from solvents and contribute in differing degrees to ozone generation. The only significant process forming O3 in the lower atmosphere is the photolysis of NO2 (reaction with sunlight), followed by the rapid reactions of the oxygen atoms formed with O2. The only significant process forming O3 in the lower atmosphere is the photolysis of NO2 (reaction with sunlight), followed by the rapid reactions of the oxygen atoms formed with O2. NO2 + hν → O(3P) + NO

[17.4.7]

O(3P) + O2 + M → O3 + M This is reversed by the rapid reaction of O3 with NO, O3 + NO → NO + O2

[17.4.8]

This reaction cycle results in a photostationary state for O3, where concentrations only depend on the amount of sunlight available, dictated by the NO2 photolysis rate (k1) and the [NO2]/[NO] concentration ratio.

[O 3 ]steady −state

=

k1 [NO 2 ] k 2 [NO ]

[17.4.9]

Because of this photostationary state, ozone levels generally rise and fall with the sun, behavior that is referred to as “diurnal.” If the above NOx cycle were the only chemical process at work, the steady-state concentrations of ozone would be relatively low. However, when VOCs such as organic solvent compounds are present, they react to form radicals that may either (1) consume NO or (2) convert NO to NO2. This additional reaction cycle combined with the above photostationary state relationship causes O3 to increase.

1194


Although many types of reactions are involved,4,13,16,17 the major processes for most VOCs can be summarized as follows: VOC + OH → RO2 + products

[17.4.10a]

RO2 + NO → NO2 + radicals

[17.4.10b]

radicals → ...→ OH + products

[17.4.10c]

products → ...→...+ CO2

[17.4.10d]

The last two pseudo-reactions given comprise many steps, and the products often include formaldehyde, carbon monoxide and organonitrates. The rate of ozone increase caused by these processes depends on the amount of VOCs present, the type of VOCs present, and the level of OH radicals and other species with which the VOCs can react. One of the major determinants of a compound’s impact on ozone is the rate of the reaction of the particular VOC with the hydroxyl radical via reaction [17.4.10a], above. The total amount of ozone formed is largely determined by the amount of VOC and NOx available. The dependence of O3 production on the initial amounts of VOC and NOx is frequently represented by means of an ozone isopleth diagram. An example of such a diagram is shown in Figure 17.4.1. The diagram is a contour plot of ozone maxima obtained from a large number of air quality model simulations using an atmospheric chemical mechanism. Initial concentrations of VOC and NOx are varied; all other variables are held constant. Notice that there is a “ridge” along a certain VOC-to-NOx ratio where the highest ozone concentrations occur at given VOC levels. This is referred to as the “optimum” VOC-to-NOx ratio. While the atmosphere is more complicated than this idealized system, important features are very similar. VOC-to-NOx ratios sufficiently low to retard ozone formation from an optimum ratio (represented in the upper left quadrant of Figure 17.4.1) can occur in central cities and in plumes immediately downwind of strong NOx sources. Rural environments tend to be characterized by fairly high VOC-to-NOx ratios because of the relatively rapid removal of NOx

Figure 17.4.1. Ozone isopleth diagram showing the dependencies of ozone on varying levels of initial VOCs and NOx. Concentrations are given in ppb. [Adapted from M.S. Bergin et al., Enc. of Env. Analysis and Remediation, 29, 3029, (1998)]


1195

from non-local sources as compared to that of VOCs, coupled with the usual absence of strong local NOx sources and the presence of natural VOC sources. In such rural environments, the formation of ozone is limited more by the absence emissions of NOx, and most ozone present was directly transported from upwind. Indeed, in most of the troposphere, except in areas of strong NOx sources, the availability of NOx governs ozone production. 17.4.3.3 Assessing solvent impacts on ozone and VOC reactivity As mentioned previously, the contribution of solvents to the VOC levels, and hence ozone formation, is significant. For example, in Los Angeles, about 25% of the VOC mass is from solvent use.1 This fraction is down from earlier years due to various controls such as using water-based paints and enclosing/controlling paint spraying operations. On the other hand, reduction in the use of CFCs as propellants has led to an increase in organic emissions from substituted compounds.2 However, the impact on ozone formation by a specific source is not directly proportional to the amount of VOC emitted by that source. A major determinant of the ozone forming potential is the reactivity of the compound or compound mixture emitted. Reactivity can be viewed as the propensity for a compound to form ozone, and this propensity varies dramatically between compounds and between environments.18,19 As seen in Table 17.4.1, ‘box’ model (single cell) simulations designed to represent summertime conditions in Los Angeles, California indicate that the amount of carbon associated with each class of compound only roughly corresponds to the amount of ozone formed from those compounds. Methane, which reacts very slowly but comprises most of the carbon, contributes little to ozone formation. Alkenes and aromatics are only a small part of the total carbon, but lead to much of the ozone formation. Table 17.4.1. Percentage of ozone production attributable to each organic. The percentages shown should be viewed as only approximate, and will depend upon local emissions characteristics. (*While not considered organic carbon, carbon monoxide acts to facilitate ozone formation similar to organic compounds.) [Adapted from F.M. Bowman and J.H. Seinfeld, J. Geophys. Res., 99, 5309, (1994) and M.S. Bergin et al., Env. Sci. Technol., 29, 3029 (1998)] Percent of carbon in each specified class

Percent of ozone due to specified organic class

carbon monoxide*

35

6

methane

40

1

aldehydes and ketones

1

3

non-methane alkanes, ~4C

8

17

non-methane alkanes, ~8C

5

16

aromatics, including toluene

3

5

aromatics, including xylenes and others

3

13

ethene

2

12

biogenic alkenes ans isoprene

1

10

other alkenes

2

17

Compound Class

1196


17.4.3.3.1 Quantification of solvent emissions on ozone formation Two methods are generally employed to quantify the role pollutants play in forming ozone: experimental and computational. Both types of estimation approaches have their limitations. In the case of physical experiments, it is difficult to fully simulate ambient conditions, so the results do not have general applicability. In the case of computational approaches, uncertainties and approximations in the model for airshed conditions, in its formulation, and in the chemical mechanism cause uncertainties in the predicted ozone impacts. For these reasons, modeling predictions and experimental measurements are used together. 17.4.3.3.1.1 Experimental analysis Experimental analysis is performed using environmental ‘smog’ chambers, either with a series of single hydrocarbons irradiated in the presence of NOx or using complex mixtures to simulate, for example, automobile exhaust emitted into characteristic urban ambient conditions. Such chambers are large reaction vessels (some with internal volumes of cubic meters), in which air and small amounts of hydrocarbons and NOx are injected, and then irradiated with real or artificial light. Both indoor and outdoor chambers are used so behaviors can be evaluated under natural radiative conditions and under controlled conditions. While these experiments18-23 clearly indicate differences in ozone formation from individual hydrocarbons, they do not represent some important physical systems of urban pollution such as the mixing processes and continuing emissions cycles. Such experiments have focused both on groups of compounds as well as specific VOCs, including solvents. A particular limitation has been studying very low vapor pressure solvents because it is difficult to get enough of the compound into the vapor phase in the chamber to appreciably change the ozone levels. Another limitation is the expense of using smog chambers to simulate a large range of conditions that might occur in the atmosphere. On the other hand, smog chambers are very powerful, if not fundamental, for developing chemical mechanisms that describe the reaction pathways that can be used in computational approaches. 17.4.3.3.1.2 Computational analysis (air quality models) Given the limitations of physical experiments to simulate atmospheric conditions, computer models have been developed to assess the impact of emissions on ozone. These models, called airshed models, are computerized representations of the atmospheric processes responsible for air pollution, and are core to air quality management.23 They have been applied in two fashions to assess how solvents affect ozone. One approach is to conduct a number of simulations with varying levels of solvent emissions.2 The second approach is to evaluate individual compounds and then calculate the incremental reactivity of solvent mixtures.19,21,24-28 Derwent and Pearson2 examined the impact of solvent emissions on ozone by simulating air parcel trajectories ending in the United Kingdom and perturbing the emissions to account for an anticipated 30% mass reduction in VOCs from solvents between 1995 and 2007. They found a small decrease in ozone-from 78 to 77 ppb in the mean peak ozone in the UK, and a 9 ppb reduction from 129 ppb outside of London. A more substantial decrease of 33 ppb from the 129 ppb peak outside of London was found from reducing non-solvent mass VOC emissions by 30% outside of the UK and 40% within the UK. This suggests that the VOC emissions from sources other than solvents have a higher average reactivity, as is discussed by McBride et al.29 While the types of simulations conducted by Derwent and Pearson2 are important to understanding the net effect of solvent emissions on ozone, there is an unanswered associated and important question, that being which specific solvents have the greatest impacts.


1197

This question is critical to assessing if one solvent leads to significantly more ozone formation than a viable substitute (or vice versa). To evaluate the contribution of individual organic compounds to ozone formation, the use of incremental reactivities (IR) was proposed,18-21 defined as the change in ozone caused by a change in the emissions of a VOC in an air pollution episode. To remove the dependence on the amount of VOC added, incremental reactivity is defined by equation [17.4.11] as the limit as the amount of VOC added approaches zero, i.e., as the derivative of ozone with respect to VOC: IRi =

∂[O 3 ]

∂[VOCi ]

[17.4.11]

Here, IRi is the incremental reactivity and the subscript i denotes the VOC being examined. This definition takes into account the effects of all aspects of the organic’s reaction mechanism and the effects of the environment where the VOC is emitted. A similar quantity is the relative reactivity,23 RRi: RRi =

Ri

[17.4.12]

N

∑F i =1

Bi

Ri

where: FB i IRi

mass fraction of compound i in the reference mixture incremental reactivity of species i (grams ozone formed per gram compound i emitted)

In this case, the incremental reactivity is normalized by the reactivities of a suite of organics, thus removing much of the environmental dependencies found when using IRs defined by [17.4.11]. This metric provides a means for directly comparing individual compounds to each other in terms of their likely impact on ozone. A number of investigators have performed calculations to quantify incremental and/or relative reactivities for various solvents and other organics23-28,30 and references therein. Those studies found very similar results for the relative reactivities of most compounds found in solvents. Figure 17.4.2 (based on references 23, 24 and 26) shows the relative reactivities for some of the more common compounds, as well as possible solvent substitutes and isoprene, a naturally emitted organic. (For a more extensive list of relative reactivities, see 19, 27 and 30.) As can be seen, even normalized compound reactivities can vary by orders of magnitude. Some compounds even exhibit “negative” reactivities, that is that their emission can lead to ozone decreases under specific conditions. In particular, negative reactivities are found most commonly when the levels of NOx are low, e.g., in non-urban locations. For example, Kahn et al.,26 found that a solvent can promote ozone formation in one area (e.g., near downtown Los Angeles), but retard ozone formation further downwind. Kahn also found that the relative reactivities of the eight different solvents studied were similar in very different locations, e.g., Los Angeles, Switzerland and Mexico City. Looking at Figure 17.4.2, it is apparent that alkenes and aromatic hydrocarbons with multiple alkyl substitutions (e.g., xylenes and tri-methyl benzene) have relatively high reactivities. Alcohols, ethers and alkanes have lower reactivities. Halogenated organics have some of the lowest reactivities, so low that they are often considered unreactive. This suggests that there are two ways to mitigate how solvents contribute to air quality problems. The more traditional method is to reduce the mass of organic solvent emissions (e.g., by using water-based paints). A second approach is to reduce the overall reactivity of the solvent

1198


Figure 17.4.2. Solvent relative reactivities based on mass of ozone formed per gram of solvent emitted into the gas phase. PCBTF is para-chlorobenzo-trifluoride, BTF is benzo-trifluoride and TBA is tertiary butyl acetate. [Adapted from M.S. Bergin et al., Env. Sci. Technol., 29, 3029 (1998) and M. Khan et al., Atmos. Env., 33, 1085 (1999)].

used, e.g., by switching to ethers, alcohols, alkanes or halogenated compounds. Solvent substitution, however, is complicated by the need to maintain product quality. 17.4.4 REGULATORY APPROACHES TO OZONE CONTROL AND SOLVENTS Historically, regulatory approaches to reducing ozone concentrations have relied reducing the mass emissions of VOCs,2,4 and this has led to stringent controls on solvents. Two factors are important in determining if an organic solvent is considered a VOC: its reactivity (discussed above) and its vapor pressure. In the U.S., traditionally, if a compound was less reactive than ethane, it was considered unreactive. Such compounds include many halogenated species and some acetates and ethers. Recently, acetone was also added as an unreactive compound. A vapor pressure threshold is also used in many areas (e.g., Europe) since it is viewed that compounds with very low vapor pressures will not be emitted rapidly into the atmosphere. It has been argued that a vapor pressure limit may not be appropriate since, given time, even lower vapor pressure compounds will have ample time to evaporate. Just recently, California is considering regulations that more fully account for the full range of reactivities that solvents possess. This is due, in part, to make it easier for manufacturers to meet stringent regulations being adopted in that state to help them meet their air quality goals. It is likely that other areas will also have to employ increasingly more stringent regulations, to both lower ozone and alleviate other environmental damage. In many countries, greater focus is now being placed on reducing NOx emissions to mitigate ozone formation. This has important ramifications for solvent use, indicating the


1199

regulatory focus is now turning from VOCs towards NOx, the other main precursor to ozone. Another imminent regulatory issue is the control of ambient fine particulate matter. While the role of solvent emissions in forming particulate matter is not well understood, studies to date do not suggest they are a major contributor. 17.4.5 SUMMARY Solvents are, and will continue to be, one of the major classes of organic compounds emitted into the atmosphere. These compounds have a wide range of air quality impacts. Accumulation of toxic compounds indoors is of concern, although outdoors the concern of toxicity is significantly less substantial due to rapid dilution. In the stratosphere, some of the halogenated solvents lead to depletion of the protective layer of ozone, while in the troposphere solvents generally lead to increased ozone levels, where it adversely affects health and the environment. The former has led to regulations of CFCs, and the latter to regulations of organic solvents. Some solvents are also considered to be precursors to the formation of secondary tropospheric pollutants other than ozone, such as particulate matter, however these relationships are currently less certain. In the aggregate, total VOC emissions from solvents in the U.S. are the second largest single source category in polluted urban areas, falling just behind motor vehicle VOC emissions both in terms of mass and urban ozone production. For now, regulations are designed to reduce the loss of ozone in the stratosphere and the formation of excess ozone in the troposphere. However, while some solvents are very reactive, others are substantially less reactive, suggesting that there is considerable opportunity to reduce urban ozone formation from solvents by utilizing substitutes with low ozone forming potentials. Currently, most regulations are targeted at reducing the mass of VOC emissions, not their relative impacts on ozone. REFERENCES 1 2 3

4 5 6 7 8 9 10

11 12 13 14 15 16

SCAQMD (South Coast Air Quality Management District). (1996). 1997 Air Quality Management Plan. November. R.G. Derwent and J.K. Pearson, Environ. Technol., 18, 1029 (1997). Climate Change 1995: The Science of Climate Change. Contribution of Working Group I to the Second Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). Ed. J.T. Houghton, et al. Cambridge University Press, Cambridge, Great Britain. 1996. NRC (National Research Council). Rethinking the Ozone Problem in Urban and Regional Air Pollution. National Academy Press, Washington, DC. (1991). OTA (U.S. Congress, Office of Technology Assessment). Catching Our Breath: Next Steps for Reducing Urban Ozone, OTA-O-412, U.S. Government Printing Office, Washington, DC. July,1989. S.M. Horvath and D.J.McKee (1994). In D.J. McKee, ed., Tropospheric Ozone: Human Health and Agricultural Impacts, CRC Press/Lewis Publishers, Boca Raton, FL, Chapter 3. M. Lippmann, Environ. Sci. & Technol., 25(12), 1954, (1991). M. Lippmann, J. of Exposure Analysis and Environmental Epidemiology, 3(1), 103, (1993). W.L.Chameides, P.S. Kasibhatla, J Yienger, and H. Levy II., Science, 264(5155), 74, (1994). D.T. Tingey, D.M. Olszyk, A.A.Herstrom, and E.H. Lee, (1994). “Effects of Ozone on Crops” in D.J. McKee, ed., Tropospheric Ozone: Human Health and Agricultural Impacts, CRC Press/Lewis Publishers, Boca Raton, FL, Chapter 6. S.B.McLaughlin and L.J. Downing, Nature, 374(6519), 252, (1995). W. Mehlhorn, Nature, 327, 417, (1989). J.H. Seinfeld and S.N. Pandis, Atmospheric Chemistry and Physic: From Air Pollution to Climate Change, New York: J. Wiley & Sons, (1998). J.A. Logan, M.J. Pather, S.C. Wofsy, and M.B. McElroy, J. Geophys. Res., 86, 7210, (1981). F.W.Lurmann and H.H. Main, Analysis of the Ambient VOC Data Collected in the Southern California Air Quality Study, final report, Contract A832-130, California Air Resources Board, Sacramento, CA., 1992. R. Atkinson, Atmos. Environ., 24A, 1, (1990).

1200

17 18 19 20 21 22

23

24 25 26 27 28 29 30


R. Atkinson, J. Phys. Chem. Ref. Data, Monograph No. 2, (1994). W.P.L. Carter, Atmos. Environ., 24A, 481, (1990). W.P.L.Carter, (1991). Development of Ozone Reactivity Scales for Volatile Organic Compounds, EPA 600/3-91-050. U.S. Environmental Protection Agency, Research Triangle Park, NC (August). W.P.L. Carter and R. Atkinson, Environ. Sci. and Technol., 23, 864, (1989). W.P.L. Carter, Atmos. Environ., 29, 2513, (1995). N.A. Kelly and P. Wang, (1996) Part I: Indoor Smog Chamber Study of Reactivity in Kelly, N.A.; Wang, P.; Japar, S.M.; Hurley, M.D.; and Wallington, T.J. (1996). Measurement of the Atmosphere Reactivity of Emissions from Gasoline and Alternative-Fueled Vehicles: Assessment of Available Methodologies, Second-Year Final Report, CRC Contract No. AQ-6-1-92 and NREL Contract No. AF-2-112961. Environmental Research Consortium, (September). M.S. Bergin, A.G. Russell, W.P.L. Carter, B.E. Croes, and J.H. Seinfeld, Ozone Control and VOC Reactivity, in the Encyclopedia of Environmental Analysis and Remediation, Ed. R.A. Meyers, J. Wiley & Sons, Inc, New York, NY. 1998. M.S. Bergin et al., Env. Sci Technol., 29, 3029 (1998) F.M. Bowman and J.H. Seinfeld, J. Geophys. Res., 99, 5309, (1994a). M. Khan et al., Atmos. Env., 33, 1085 (1999). W.P.L. Carter, J. Air and Waste Mgmt. Assoc., 44, 881, (1994a). R.G. Derwent, and M.E. Jenkin, Atmos. Environ., 25(A):1661-1673 (1991). S. B. McBride et al., Env. Sci. Technol., 31, 238a, (1997) http://www.cert.ucr.edu/~carter/bycarter.htm

18

Concentration of Solvents in Various Industrial Environments 18.1 MEASUREMENT AND ESTIMATION OF SOLVENTS EMISSION AND ODOR Margot Scheithauer Institut für Holztechnologie Dresden, Germany

18.1.1 DEFINITION “SOLVENT” AND “VOLATILE ORGANIC COMPOUNDS” (VOC) Solvents are generally understood to be substances that can physically dissolve other substances, more narrowly they are inorganic and organic liquids able to dissolve other gaseous, liquid, or solid substances. A qualifier for the suitability as a solvent is that, during the solution, neither the solvent nor the dissolved substance undergoes chemical change, i.e., the components of the solution may be recovered in their original form by physical separation processes, such as distillation, crystallization, sublimation, evaporation, adsorption. From a chemical point of view, solvents or volatile organic compounds, VOCs, vary widely. They are often classified using their boiling point, and vapor pressure. These properties do not define their suitability as solvents. In the context of VOCs, only organic solvents are of relevance. Therefore, in the following discussion, only organic solvents will be discussed. Definition of solvent according to Council Directive 1999/13/EC of 11 March 19991 Organic solvent shall mean any VOC which is used alone or in combination with other agents, and without undergoing a chemical change, to dissolve raw materials, products or waste materials, or is used as a cleaning agent to dissolve contaminants, or as a dissolver, or as a dispersion medium, or as a viscosity adjuster, or as a surface tension adjuster, or a plasticizer, or as a preservative. Definition of solvent according to ISO/DIS 4618-4: 19992 Paints and varnishes - terms and definition for coating materials: Solvent: A single liquid or blends of liquids, volatile under specified drying conditions, and in which the binder is completely soluble. For certain application purposes, solvents are defined more specifically: Technische Regeln für Gefahrstoffe TRGS 610 (Technical Regulations for Hazardous Substances) as applied to high solvent-containing primers and flooring glues the definition is:

1202

Margot Scheithauer

Solvents are volatile organic compounds as well as mixtures thereof at a boiling point < 200°C, which under normal conditions (20°C and 1013 hPa) are liquid and are applied for dissolving and diluting other substances without undergoing chemical change.3 As can be seen, there are clear differences in the way VOCs are defined.4 The most general, and hence least disputable, one is: Definition of volatile organic compounds according to DIN ISO 11890/1,2,5,6 Solvent is generally any organic liquid and/or any organic solid substance, which evaporates by itself under prevailing conditions (temperature and pressure). Definition of VOC according to Council Directive 1999/13/EC1 Volatile organic compound, VOC, shall mean any organic compound having at 293.15 K a vapor pressure of 0.01 kPa or more, or having a corresponding volatility under the particular conditions of use. For the purpose of this Directive, the fraction of creosote, which exceeds this value of vapor pressure at 293.15 K shall be considered as a VOC. In the Lösemittelverordnung Österreichs (Austrian Regulations on Solvents) of 1995,7 which on this issue corresponds to the österreichische LackieranlagenVerordnung (Austrian Regulation on Varnishing Plants) of 1995,8 VOCs have a maximum boiling point of 200°C. Hence, the volatile compounds at a boiling point > 200°C are not included. These comprise, e.g., some film forming media, such as butyldiglycol, butyldiglycolacetate, and texanol. Similarly, reactive solvents are excluded from this regulation. Definitions of VOCs in the USA according to ASTM D 3960-19 Volatile Organic Compound (VOC), means any compound of carbon, excluding carbon monoxide, carbon dioxide, carbonic acid, metallic carbides or carbonates, and ammonium carbonate, which participates in atmospheric photochemical reactions. This includes any such organic compound other than the following, which have been determined to have negligible photochemical reactivity: Methane; ethane; methylene chloride (dichloromethane); 1.1.1-trichloroethane (methyl chloroform); 1.1.1-trichloro-2.2.2-trifluorethane (CFC-113), furthermore cyclic, branched or linear completely methylated siloxanes: acetone: and perfluorocarbon compounds. The regulatory definition under the control of the U.S. EPA may change. Certain organic compounds that may be released under the specified bake conditions are not classified as VOC as they do not participate in atmospheric photochemical reactions. Such non-photochemically active compounds are referred to as exempt volatile compounds in the practice. An example of an exempt compound in the USA according to the EPA is tertiary-butyl-acetate (TBA, B.p. of 98 °C). It is described as a substance not harmful in air.10 Substances contained in air are differentiated in their mixtures according to their volatility:11 Abbreviation

Compound class

Range of boiling point in °C

VVOC

very volatile organic compounds

< 0 up to 50 ÷ 100

VOC

volatile organic compounds

50 ÷ 100 up to 250 ÷ 260

SVOC

semi-volatile organic compounds

250 ÷ 260 up to 380 ÷ 500

TVOC means “total volatile organic compounds”.

18.1 Measurement and estimation of solvents emission and odor

1203

18.1.2 REVIEW OF SOURCES OF SOLVENT EMISSIONS 18.1.2.1 Causes for emissions The basic human needs include: eating, drinking and breathing. In a 60-year lifetime, human being takes up about 30 t of food, 60 t of drink and 300 t of air.12 Thus air pollution may have a large influence on human health. In comparison with outside air and the ambient air within public transportation vehicles and terminals, room air quality is of the greatest interest, since the population in North American cities stays indoors more than 93% time according to findings by Szalai13 and more than 91% according to Chapin,14 of this time > 67 % is spent in living space and about 4-5 % in the workplace. Sources for volatile substances in indoor air are diverse and originate, apart from that brought in with outside air and produced by the living occupants (people, pets) and their activities (e.g., smoking), from the materials themselves which have been applied for building and furnishing the room. These include: • building materials • glues • floor covering • wallpaper • internal decorative textiles • paints • furniture, upholstered furniture, etc. These materials may contain volatile organic, but also inorganic, compounds, which are evaporated during use. The emission of these volatile compounds is influenced by a large number of factors, such as: • chemical/physical structure of the material (thickness, surface structure) • volatility, polarity of the volatile compounds • room temperature and humidity • room load, i.e., the ratio of the area of emitting surface and the volume of room air • ventilation, rate of air flow • load of the external air entering (dust, substances) The residual monomers from plastics, reactive products (e.g., formaldehyde), degradation products, flame protection media, softeners etc. may be emitted from building materials. 18.1.2.2 Emissions of VOCs from varnishes and paints A substantial part of emitted materials are solvents. They preferably originate from varnishes, paints and glues. Table 18.1.1 shows a survey of the main types of varnishes referring to their average solvent content. Table 18.1.1. Solvent share in various types of varnishes Material

Solvent content, wt%

Stain

~95

Cellulose nitrate varnish

~75

PUR varnish Polyester varnish UV roller varnish based on acrylate

35-70 35 2-10

1204

Margot Scheithauer

Despite the introduction of water based and powder varnishes, cellulose nitrate varnishes with their high solvent content, are still widely used due to their easy application and their low prices. The current varnish application still requires the use of solvents. The chosen coating procedure determines the necessary processing viscosity, which may be adjusted in different ways: • Thinning the binding agent/varnish with a solvent consisting of one or more organic compounds without altering the binding agent chemically. Nowadays, the solvent can partly or almost totally be replaced by water. • Thinning the binding agent with a reactive thinner, i.e., one or more monomers/oligomers included in the system, which react with the binding agent during the hardening process and is built into a molecule. According to definitions, they do not have characteristics of solvents. They partly take over the solvent function while they remain in the liquid state. • Temperature increase of the coating system, which limits the required amount of solvents. • Liquid application of a melted solid system Profiled construction components form parts of complicated shapes, which must rely on the spraying of coatings for decorative and protective purposes. The spraying requires low-viscosity material. Viscosity reduction is usually achieved by adding solvents but sometimes by increasing the material temperature. Parts with large flat surfaces however, may be coated by roller-coating, which tolerates higher viscosity materials. After fulfilling their function as solvents, they are expected to completely evaporate from the varnish system. Emission of solvents from a varnish system occurs in the course of the “life-cycle” of a varnish in several different locations as the following survey shows (Figure 18.1.1). As early as in the production stage of the varnish system, solvent emissions contaminate the ambient air of the factory. About 90 % of the solvents contained in the varnish system evaporates during its application which affects the air quality in the workplaces. These emissions are in the milligram per cubic meter of air range. The thinning effect decreases the solvent concentration in the ambient air at the varnisher’s workplace into the microgram to nanogram per cubic meter.

Figure 18.1.1. Emission of VOCs in the life-cycle of a varnish.


1205

About 10 % of the solvent remains in the varnish and it is emitted, depending on the type of solvent and varnish, with variable rate on their way to the consumers and in their living spaces. According to the European Commission, the furniture industry alone used more than 185,000 tons of solvents/year in 1994.15 Not all the VOCs enter the air, however, since remedial efforts, such as absorption and solvent reclaim, after-burning of solvents, the use of bio-filters, result in emission reduction. Nevertheless, it is still evident that limitation of solvents, VOCs, reduction in varnishes and coatings may help in reducing emissions. 18.1.2.3 VOC emissions from emulsion paints Although emulsion paints and plasters contain less than 2% VOCs these are produced in such large quantities (3.5 Million t in 1995)16 and applied over such an extensive surface that 100,000 t of VOCs are emitted from these materials per year. These enter the environment in an uncontrolled way. Prior to an interior application of the coating, there is no VOC in room atmosphere, unless sources already exist in the room. Then the load - from the beginning of the application - increases very rapidly. After the application, the VOC concentration in the ambient air decreases exponentially. If these emissions are to be monitored it is desirable to have methods available to record VOC concentrations typical of emitting stages.17 18.1.3 MEASURING OF VOC-CONTENT IN PAINTS AND VARNISHES 18.1.3.1 Definition of low-emissive coating materials In order to limit the sources of solvent emission in ambient air in the most effective manner, primary goal must be to reduce the VOC content in coating materials. In Germany, there is a draft for the Varnishing Plant Regulation18 - regulating the application of low-emissive varnishes, which will require that coating materials must initially not contain more than 420 g/l VOC. Four years after the introduction of the regulation this is reduced to 250 g/l VOC. The “Jury Umweltzeichen” (Environmental Trade Mark), which, since 1999, rewards certain products with the environmental trade mark goes even further: For flat, plain materials (e.g., interior doors, panels, floors with varnished surfaces, pre-fabricated parquet) the coating systems applied must not exceed the VOC content = max. 250 g/l.19 Varnishing plants are exempt if they employ waste gas purification facilities corresponding to the EU VOC Guideline.20 The draft of the Varnishing Plant Regulation follows the US method according to ASTM D 3960-121 in determining the VOC content. 18.1.3.2 Determination of the VOC content according to ASTM D 3960-1 ASTM D 3960-1 is the method used in the US to determine the VOC content of coating materials. The VOC value is calculated as follows: VOC value [g / l ] =

mass of volatile components [g ] − mass of water [g ] volume of varnish [l ] − volume of water [l ]

[18.1.1]

For varnishes exclusively containing solvents, this formula results in high values. This calculation is, however, of dubious value for water-containing varnishes. The unit g/l does

1206

Margot Scheithauer

not refer to 1 liter of an actual varnish, but to a hypothetical product, which only expresses the binding agent in liters. This is demonstrated in Example 1, which will then also be calculated according to other methods. Example 1: Water spray varnish: 30% solid matter content (binding agent and pigments) 10% solvent content 60% water content varnish density: 1.0 g/cm3 VOC (g/l) = (700 g - 600 g)/(1 l - 0.6 l) = 250 g/l 18.1.3.3 Determination of the VOC content according to ISO/DIS 11 890/15 and 26 18.1.3.3.1 VOC content > 15% The calculation according to ISO/DIS 11 890/1 gives a more logical result. VOCs are given here as a mass share in % of the ready-to-use varnish. At a VOC content > 15 % (of the mass share), the determination is made by the differential approach: VOC [% of the mass share] = 100 - nfA - mw

[18.1.2]

where: nfA mw

non-volatile parts determined by ISO 325122 (1 h/110°C) mass as water in % of the mass parts, determined according to ISO 76023 (after Karl-Fischer)

Many more recently available techniques for water determination are not sufficiently common and thus are not considered. Example 2: Water solvent stain: 5% solid matter content = non-volatile parts 25% solvent 70% water share VOC (mass %) = 100 - 5 - 70 = 25 % For density ≠ 1 VOC [g/l] = 10(100 - nfA - mw) ρs where: ρs

varnish density in g/ml at 23°C

In this case, the VOC content refers to 1 l of actual varnish. Example 3: Solvent stain: 5% solid matter content 95% solvent share VOC (g/l) = (100 - 5 - 0) × 0.9 = 855 g/l Example 4: Solvent/water stain: 5% solid matter content 70% solvent 25% water share

[18.1.3]


1207

Density: 0.9 g/cm3 VOC (g/l) = (100 - 5 - 25) × 0.9 × 10 = 630 g/l DIN ISO 11 890 also allows the calculation to be made using the method prescribed by ASTM D 3960: VOClw [g / l ] = 1000

100 − nfAmw ρ mw s 100 − ρ s ρw

[18.1.4]

where: ρs nfA mw ρw

varnish density in g/ml at 23°C non-volatile parts determined by ISO 325122 (1 h/110°C) mass as water in % mass parts, determined acc. to ISO 76023 (after Karl-Fischer) density, in g/ml, of water at 23°C (ρw = 0.997537 g/ml at 23°C)

Example 1 Water spray varnish: 30% solid matter content 10% solvent 60% water content varnish density 1.0 g/cm3 water density 1.0 g/cm3 VOClw = [(100 - 30 - 60)/(100 - 1.0 × 60)] × 1.0 × 1000 = 250 g/l Example 1: Water spray varnish but with a density of 0.9 g/cm3 VOClw = [(100 - 3 - 60)/(100 - 0.9 × 60)] × 0.9 × 1000 = 196 g/l Or another example: Example 5: Water varnish primer: 15% solid matter content 10% solvent 75% water varnish density 1.0 g/cm3 VOClw = [(100 - 15 -75)/(100 - 1.0 × 75)] × 1.0 × 1000 = 400 g/l The calculation permits the deduction of exempt compounds, which are defined as organic compounds which do not participate in photochemical reactions in the atmosphere. i =n

VOClwe [g / l ] = 1000

100 − nfA − mw ∑ m eci i =1

i =n m m 100 − ρ s w − ρ s ∑ eci ρw i =1 ρ eci

ρs

[18.1.5]

where: VOClwe nfA mw meci ρs

VOC content, in g/l, contained in the product in its ready-to-use condition minus water and minus exempt compounds content of non-solvent parts, as a mass share in % water content, as mass share in % content, as mass share in %, of the exempt compound i density, in g/ml, of sample at 23°C

1208

Margot Scheithauer ρw ρeci 1000

density, in g/ml, of water at 23°C (ρw = 0.997537 g/ml at 23°C); density, in g/ml, of the exempt compound i conversion factor × 1000 in g/l

Example 4: Exempt compounds are propanol and ethanol Solvent water stain 5% solid matter content 20% propanol (density 0.8 g/cm3) (density 0.9 cm3) 10% ethanol (density 0.79 g/cm3) 40% other solvent 25% water VOClwe = [(100-5-25-(10))/{100-0.9(25/1.0)-0.9[(20/0.8)+(10/0.79)]}]× 0.9×1000=825 g/l Calculated according to DIN ISO 11 890-1, for the same water stain applies (example 4) after approach 1: 70% after approach 2: 630 g/l (propanol and ethanol are treated as VOC) 18.1.3.3.2 VOC content > 0.1 and < 15 % For VOC content > 0,1 (by mass) and < 15 % the determination is made using gas chromatography (ISO/DIS 11 890-2).6 A hot or the cold injection is used depending on the sample properties. After assessing the peak areas, the quantitative assessment and evaluation of the VOC content in the product in its ready-to-use condition, the calculation is made in the simplest case according to: i =n

VOC [g / l ] = ∑ m i ρ s 1000

[18.1.6]

i =1

where: VOC mi ρs 1000

VOC content, in g/l, of the ready-to-use product; mass, in g, of compound i in 1 g of the sample density, in g/ml, of the sample at 23°C conversion factor × 1000 in g/l.

Table 18.1.2 shows how the application of the various formulae affects the result of calculation. The VOC contents, calculated after ISO/DIS 11 890, are credible for Examples 2, 3, 4. Values calculated according to ASTM, using ratio solvent - water in Examples 2 and 4, are almost the same as VOC contents calculated according to ISO/DIS. Only by increasing the solid matter content, the VOC content can be reduced in examples calculated according to ASTM (cf. Examples 5 and 1), while an increase in the water content, in spite of the same solvent share, leads to an increase of the ASTM value, because a higher water content increases the sum of the denominator and consequently increases the VOC content. 18.1.3.4 Determination of VOC-content in water-thinnable emulsion paints (in-can VOC) DIN 55 64924 describes an approach, which, via the so-called total evaporation method, assesses the content of volatile organic compounds in water-thinnable emulsion paints (in-can VOC).


1209

Table 18.1.2. Examples for the VOC content calculated by various methods VOC content according to Example

Coating material wt%

Non-vola- Solvent Water Density tile parts fraction fraction 3 g/cm wt% wt% wt%

ASTM D 3960 g/l

DIN ISO 11 890/1 %

g/l

1

Water spray varnish

30

10

60

1.000 0.900

250 196

10 10

100 90

2

Water/solvent stain

5

25

70

1.000

833

25

250

3

Solvent stain

5

95

0

0.900

855

95

855

4

Solvent/water stain

5

70

25

0.900

839

70

630

5

Water varnish primer

15

10

75

1.000

400

10

100

Measuring principle: The VOCs are totally evaporated from a very small amount (a few milliliters) of the thinned original sample by means of a head-space injector and subsequent gas chromatographic analysis. All components, whose retention time is lower than the retention time of tetradecane (boiling point 252.6°C) are included as VOCs. VOC content calculation: VOC [mg / kg ] = 1000

mVOC Eρ

[18.1.7]

where: mVOC Eρ

mass in mg VOC, related to originally weighted-in quantity of the original sample Eρ originally weighted-in quantity in g of the original sample

VOC[g / l ] =

VOC[ mg / kg ] ρ df 100

[18.1.8]

where: ρdf

density in g/ml of the original sample (emulsion paint)

18.1.4 MEASUREMENT OF SOLVENT EMISSIONS IN INDUSTRIAL PLANTS 18.1.4.1 Plant requirements In spite of considerable reductions in the use of organic solvents, the worldwide-adopted ozone values in the troposphere are not being attained. It is crucial to apply consistent measures worldwide to reduce solvent emissions. In Europe, for example, the Council Directive 1999/13/EG1 “On limiting emissions of volatile organic compounds, due to the use of organic solvents in certain activities and installations” [EU-VOC-Richtlinie] came into force on March 11, 1999. It has to be transferred into national laws within two years. The application area refers to activities in applying solvents, as enlisted in Appendix IIA of the EU-VOC-Richtlinie (Table 18.1.3).

1210

Margot Scheithauer

Table 18.1.3 Threshold and emission controls d

c a

Activity

Web offset printing (>15)

b

Threshold

15-25

100

30(1)

>25

20

30(1)

Publication rotogravure (>25) Other rotogravure, flexography, rotary screen, printing, laminating or varnishing units (>15) rotary screen printing on textile/cardboard (>30) Surface (1) cleaning (>1)

e

Fugitive Total emission emission values, % Exist. New Exist. New

Emission limit, mgC/Nm³

75

15-25 >25 >30(1)

10

100 100 100

Special provisions (1)

Solvent residue in finished product is not to be considered as part of fugitive emissions

15

(1)

25 20 20

Threshold for rotary screen printing on textile and on cardboard

(1)

1-5 >5

(2)

20 20(2)

Using compounds specified (2) in Article 5(6) and (8); Limit refers to mass of compounds in mg/Nm³, and not to total carbon

15 10

(1)

Other surface cleaning (>2)

2-10 >10

75(1) 75(1)

20(1) 15(1)

Installations which demonstrate to the competent authority that the average organic solvent content of all cleaning materials used does not exceed 30 wt% are exempt from application of these values (1)

Vehicle coating (0,5

(1)

50

Compliance in accordance with Article 9(3) should be demonstrated based on 15 minute average measurements

25

(1)

Coil coating (>25)

50(1)

5

10

For installations which use techniques which allow reuse of recovered solvents, the emission limit shall be 150


d

c a

Activity

b

Threshold


1211

e



Other coating, including metal, plastic, textile(5), fabric, film and paper coating

5-15 >15

100(1)(4) 50/75(2)(3)(4)

Emission limit value applies to coating application and drying processes operated under (2) contained conditions. The first emission limit value applies to drying processes, the second to coating application (3) processes. For textile coating installations which use techniques which allow reuse of recovered solvents, the emission limit applied to coating application and drying processes taken (4) together shall be 150. Coating activities which cannot be applied under contained conditions (such as shipbuilding, aircraft painting) may be exempted from these values, in accordance with Article (5) 5(3)(b). Rotary screen printing on textile is covered by activity No 3

20(4) 20(4)

(>5)

Winding wire coating (>5)

10 g/kg(1) 5 g/kg(2)

(1)Applies to installations where average diameter of wire (2) ≤0.1mm. Applies to all other installations (1)

Coating of wooden surfaces (>15)

15-25 >25

100(1) 50/70(2)

25 20

Emission limit applies to coating application and drying processes operated under con(2) tained conditions. The first value applies to drying processes, the second to coating application processes

1212

Margot Scheithauer

d

c a

Activity

b

Threshold


e



20 g/kg(1)(2)(3)

Expressed in mass of solvent emitted per kilogram of product (2) cleaned and dried. The emission limit in Article 5(8) does (3) not apply to this sector. The following exemption refers only to Greece: the total emission limit value does not apply, for a period of 12 years after the date on which this Directive is brought into effect, to existing installations located in remote areas and/or islands, with a population of no more than 2000 permanent inhabitants where the use of advanced technology equipment is not economically feasible

11 kg/m³

Does not apply to impregnation with creosote

85 g/m² 75 g/m² 150 g/m²

Emission limits are expressed in grams of solvent emitted per (1) m² of product produced. For leather coating activities in furnishing and particular leather goods used as small consumer goods like bags, belts, wallets, etc.

Footwear manufacture (>5)

25 g per pair

Total emission limit values are expressed in grams of solvent emitted per pair of complete footwear produced

Wood and plastic lamination (>5)

30 g/m²

Dry cleaning

Wood impregnations (>25)

Coating of leather (>10)

(1)

100(1)

45

10-25 >25 (>10)(1)

(1)

Adhesive coating (>5)

5-15 >15

(1)

50 50(1)

25 20

If techniques are used which allow reuse of recovered solvent, the emission limit value in waste gases shall be 150


d

c a

Activity

Manufacture of coating preparations, varnishes, inks and adhesives (>100)

Threshold


100-1000 >1000

150 150

b

1213

e


5% of solvent input

5 3

Special provisions

The fugitive emission value does not include solvent sold as part of a coatings preparation in a sealed container (1)

Rubber conversion (>15)

20(1)

25(2)

25% of solvent input

If techniques are used which allow reuse of recovered solvent, the emission limit value in (2) waste gases shall be 150. The fugitive emission value does not include solvent sold as part of products or preparations in a sealed container

f see below the table

Total emission limit values for installations processing individual batches of seeds and other vegetable matter should be set by the competent authority on a case-by-case basis, applying the best available (2) Applies to all techniques. fractionation processes excluding de-gumming (the removal (3) of gums from the oil). Applies to de-gumming

(1)

Vegetable oil and animal fat extraction and vegetable oil refining activities (>10)

(1)

Manufacturing of pharmaceutical products (>50)

20(1)

5(2)

15(2)

5% 15% of solvent input

If techniques are used which allow reuse of recovered solvent, the emission limit value in (2) waste gases shall be 150. The fugitive emission limit value does not include solvent sold as part of products or preparations in a sealed container

a (solvent consumption threshold in tonnes/year); b(solvent consumption threshold in tonnes/year); cvalues in waste gases; d(percentage of solvent input); elimit values; fAnimal fat: 1.5 kg/tonne, Castor: 3 kg/tonne, Rape seed: 1 kg/tonne, Sunflower seed: 1 kg/tonne, Soya beans (normal crush): 0.8 kg/tonne, Soya beans (white flakes): 1.2 kg/tonne, Other seeds and other vegetable matter: 3 kg/tonne(1) 1.5 kg/tonne(2) 4 kg/tonne(3)

The quoted threshold levels for the solvent consumption in tonnes/year is of decisive importance as to whether a plant falls under this directive. This value changes depending on the technical feasibility within the industry. For example, the threshold level for solvent consumption is for illustration-gravure-printing > 25 tonnes/year, for wood coating > 15 tonnes/year.

1214

Margot Scheithauer

An analysis of the activity requires data on solvent consumption as calculated from the amount of solvent purchased both as pure solvent and included in solvent containing materials less the amount of solvent retained and/or contained in waste. The calculated difference includes all emissions including diffuse emissions such as the solvent loss from drying racks or solvent initially retained by pained, varnished, printed or dry cleaned articles. These measurements are essential if the statutes are to be enforced and emissions are to be effectively reduced. Emissive limit values are stipulated for plants covered under this directive for exhaust-gases in mgC/Nm³ as well as limits for diffuse emissions in % solvents input. If these data are not available, total emission limit values are used. Definition of “Emission Limit Value”: The “Emission Limit Value” is understood as the mass of volatile organic compounds, the concentration, the percentage and/or the amount of emission - ascertained under normal conditions - expressed in certain specific parameters, which in one or several time periods must not be exceeded [EU-VOC-Richtlinie].1 Plant measurement must be made to provide data for analyzing the actual situation as well as for being able to prove the emission reduction which business must attain. 18.1.4.2 The determination of the total carbon content in mg C/Nm³ 18.1.4.2.1 Flame ionization detector (FID) The total C/Nm³ is assessed according to the Guideline VDI 3481/page 325 by means of a flame ionization detector (FID). This device is the component of a mobile device for random sample tests or a continuously measuring device for total carbon concentration measurement in an exhaust-gas flow. This approach measures the total organic substance in an exhaust gas. Should the composition of the solvents contained in the exhaust air not be known, their concentration may be quoted in carbon equivalents as mgC/m³. Measuring principle: An FID detects ionized organically compounded C atoms in a hydrogen flame. The ion flow developed in the induced electrical field is electrically amplified and measured. The ion flow arising when burning carbon compounds is proportional to the mass of carbon atoms exposed to the flame per time unit. The detection of organic compounds with heteroatoms, e.g., N, O, S, Cl, is generally less sensitive. The calibration of the device is done in most cases with propane as the test gas. Procedural data (measuring ranges, proof limits, etc.) by FIDs of various types are compiled in a table in the Guidelines VDI. Figure 18.1.2 shows a measuring arrangement for assessing emissions from a drying plant by means of an FID. 18.1.4.2.2 Silica gel approach The determination of the total carbon concentration in an exhaust gas by means of the silica gel approach is effected according to the Guideline VDI 3481, p. 2.26 A partial flow of the exhaust air to be tested is guided through a sorption pipe filled with silica Figure 18.1.2. Emission measurement by FID. gel. The organic compounds are absorbed


1215

by the silica gel. Subsequently, the organic compounds are desorbed in an oxygen flow at an increased temperature and burnt to carbon dioxide (CO2), which is determined quantitatively. This is a discontinuous method. The data reflect concentration in the measured time intervals and the causes of deviations are difficult to ascertain. 18.1.4.3 Qualitative and quantitative assessment of individual components in the exhaust-gas While the methods mentioned under 18.1.4.2 reflect the total carbon concentration as a summary parameter, it is still necessary to know concentrations of the individual solvents to assess the plant emission. 18.1.4.3.1 Indicator tubes For many solvents and other substances directly indicating detector tubes are available.27 They are used for rapid assessment of emission. Since they often do not react specifically with a certain compound alone, errors have to be considered due to cross-sensitivity. Detector tubes contain sorption-active agents. A gas flow of a defined rate is fed through and the substance in the gas flow produces a color reaction with the agent in the tube. The color intensity is checked against the scale on the test tube and the ppm value can be registered. Figure 18.1.3 shows such a tube for determining ethyl acetate.27 The color change in this case is from yellow to light green. Reaction principle: CH3-COOC2H5 + CrVI → CrIII + various oxidation products This reaction is not substance-specific. Other alcohols, at varying cross-sensitivity as well as methyl ethyl ketone, are reactive as well. 18.1.4.3.2 Quantitative solvent determination in exhaust gas of Figure 18.1.3. Indicaplants by means of gas-chromatography tion tube for ethyl aceA defined amount of exhaust gas is fed into an adsorption medium (actate 500/a-D. tivated carbon, thenax, etc.) and its concentration is enriched. There is a subsequent extraction by means of a solvent. The mixture is then injected into a gas chromatograph. The individual components are determined according to Guideline VDI 34 82, page 1.28 Measuring principle: A heatable capillary coated on the inside with various materials (polyester or silicon resins, silica gel) is used as a separating unit. The gaseous substance mixture is transported by means of an inert gas (nitrogen, argon, etc.). The individual solvents remain in the column for different time periods depending on their polarity. After leaving the column solvent is detected by sensitive physical methods. By combining the highly resolving capillary gas chromatography with a mass spectrometer even unknown substances may be identified in the mixtures. The quantitative evaluation of the gas chromatograms is automatically done with appropriate software. 1st Example: Emission measurement in an industrial varnishing plant The measurement was done in a furniture company, which applies different varnish systems on several varnishing lines.29 For preliminary investigations, four chimneys were selected for the measurement as typical sources of emission.

1216

Margot Scheithauer

Based on the analysis of the varnish materials and measurements performed over extended period of time, it is certain that the materials and concentrations processed during the measuring gave a representative cross-section of the production activity. Table 18.1.4 shows a summary of the varnish and solvent amounts processed per day, which were emitted from chimneys 1 to 4. Table 18.1.4. Summary of the varnish and solvent amounts processed in the 1st example Measuring point chimney

Varnish

Solvent share, % 34 76 80 80 80

10 - 15

1

1 2 3 3 4

2

5 6

56 34

50 - 60 10

7

38

30

Varnish processed, l/day

Total solvent, l/day

30 - 35 5 5

~40 - 45

37

3

12.5

4

6

34

3

1 8 6

34 52.88 34

~240 120-130 50

~170

At the chimneys, the following exhaust air parameters were established: Chimney diameter mm Exhaust air flow rate m/s Exhaust air temperature °C Humidity % Volume flow (operation) m³/h Volume flow Nm³/h Heat flow MW Table 18.1.5 contains a summary of the established emission values expressed in mg C total/Nm³ as well as the mass flow g C/h.

1

10.35 - 11.20 11.45 - 12.00 13.45 - 14.35

35 200

112.6 96.5 128.6

3 964 3 397 4 527

93 87

Mass flow, TGE/m³

Odorant concentration(1), GE/m³

Mass flow, gC/h

Carbon concentration, mgC/Nm³

Volume flow, Nm³/h

Time

Measuring point chimney

Table 18.1.5. Measuring results for total carbon concentration at different measuring points

3 519 3 292

(1)

Mass flow, TGE/m³

Odorant concentration(1), GE/m³

1217

Mass flow, gC/h

Carbon concentration, mgC/Nm³

Volume flow, Nm³/h

Time

Measuring point chimney


2

11.25 - 12.00 12.45 - 13.10 13.10 - 14.30

39 240

72.4 56.3 56.3

2 841 2 209 2 209

91 87 93

4 151 3 969 4 242

3

13.50 - 14.40 09.25 - 09.55 09.55 - 10.25

33 900

48.2 56.3 64.3

1 634 1 908 2 180

100 115

3 747 4 309

4

09.15 - 09.45 09.45 - 10.15 10.15 - 10.45

108 650

112.5 152.8 160.8

12 223 16 602 17 471

98 213

12 837 27 900

The odorant concentration does not refer to the times stated.

The EU VOC Directive for plants for wood coating with a solvent consumption of 15 25 t/year requires an emission limit of 100 mg C/Nm³. In the example, this is only met at chimneys 2 and 3. As a parallel to these measurements, using activated carbon tubes and subsequent gas chromatographic evaluation it was determined that the main components of VOCs were acetone, ethyl acetate, toluene, butyl acetate, xylene, ethylbenzene (Table 18.1.6). Butyl acetate and ethyl acetate were the main components in the respective solvent mixtures. Table 18.1.6. Measured results for individual components at different measuring points Acetone MPC*

a

Ethyl acetate

Conc

MF

b

1

6.5 4.5

2

a

Conc

MF

b

229 158

15.5 11.0

11.0 4.5

432 176

3

-

4

9.0 5.5

Toluene a

Butyl acetate

Conc

MF

b

546 387

9.0 5.5

24.5 9.0

961 353

-

4.5 15.5

978 597

20.0 27.0

a

Ethyl benzene a

b

Xylene

Conc

MF

b

Conc

MF

Conca

MFb

317 194

15.5 11.0

545 387

1.0 1.0

352 352

4.2 2.5

158 88

14.5 6.5

569 255

24.5 14.5

961 569

1.0 -

39 -

6.5 -

255 -

152 525

2.0 5.5

68 186

3.5 8.0

118 271

0.5 1.0

17 34

2.0 5.5

68 186

2 173 2 933

9.0 18.0

978 1 956

17.0 37.0

1 847 4 020

2.0 3.5

217 380

8.0 17.0

869 1 847

*MPC = Measuring point chimney; aconcentration in mg/Nm³; bMF = mass flow in g/h

2nd Example: Emissions measured in a spray-room The usual technique of measuring VOC emissions in plants by means of the FID according to VDI 3481/page 3 has the shortcoming that individual components in the gas mixture cannot be measured separately, but only determined as total carbon. Also the device needs to be placed in the immediate vicinity of the measuring point and requires frequent calibration.

1218

Figure 18.1.4. Photoacoustic Multigas Monitor 1312.

Margot Scheithauer

A portable multi-gas monitor 1312 [Bruel & Kjaer]30 weighing 9 kg (Figure 18.1.4) enables sampling in six places up to 50 m away from the monitor. It can determine at each measuring point the concentration of up to a maximum 5 components as well as the humidity. This is done by a multi-point sampler with a dosing apparatus controlled by system software. The device needs calibration only four times a year. Beyond that, the device may also be used for determining the total C concentration.

Measuring principle (Figure 18.1.5) 1. The “new” air sample is hermetically sealed in the analysis cell by closing the inlet and outlet valves. 2. Light from an infrared light source is reflected by a mirror, passed through a mechanical chopper, which pulsates it, and then passes through one of the optical filters in the filter carousel. 3. The light transmitted by the optical filter is selectively absorbed by the gas monitored, causing the temperature of the gas to increase. Because the light is pulsating, the gas temperature increases and decreases, causing an equivalent increase and decrease in the pressure of the gas (an acoustic signal) in the closed cell. 4. Two microphones mounted in the cell wall measure this acoustic signal, which is directly proportional to the concentration of the monitored gas present in the cell. 5. The filter carousel turns so that light is transmitted through the next optical filter, and the new signal is measured. The number of times this step is repeated depends on the number of gases being measured.

Figure 18.1.5. Measurement principle of Photoacoustic Multigas Monitor Type 1312.


1219

Figure 18.1.6. Maximum concentration of n-butyl acetate in the spray-room of a joiner’s with windows and door closed. A - Spray wall, B - Rack track for shelving the varnished parts.

The response time is only 13 s for a single gas or water-vapor, or approx. 40 s if 5 gases and water-vapor are measured. The measurements are quasi-continuous. When measuring a single gas, the measuring value is updated every 30 s, with five gases and humidity the update time is 105 s. The measured values are stored, statistically evaluated and numerical or graphical output can be obtained either immediately or on demand. Such devices make the emission measurement effort much easier than with the older, less automated equipment. Figure 18.1.6 shows the results of emission investigations in the spray-room of a joiner’s with and without ventilation.31 Windows and doors were closed. The varnisher was standing in front of a spray-wall, opposite to him was placed a platform rack truck for shelving the varnished parts. Those areas in the joiner’s, which showed roughly the same emission values of n-butyl acetate, are marked on each drawing. The emission load in the room was decreased by the technical ventilation to about a tenth of the values. 18.1.5 “ODOR” DEFINITION One of the senses of the living organism is the sense of smell. Smell is caused by a substance whose effect is largely dependent on its chemical structure. The odorous substances perceived by human are suspended in the air as volatile substances. The degree of response to a substance depends on its vapor pressure, thus, in turn on its concentration in the air and its molecular weight. An odorant also has to be sufficiently water and fat-soluble, which enables it to interact with the olfactory receptors.

1220

Margot Scheithauer

Whether an odorous impression is considered pleasant or unpleasant is largely determined by the functional groups of the chemical compounds (Table 18.1.7). Table 18.1.7. Scent qualities of various chemical compounds in relation to their functional groups33,34 Odorous impression

Functional group

Substance group

Pleasant

-OH -OR -CHO -COR -COOR -CN -NO2

Alcohols Ethers Aldehydes Ketones Esters Cyanogen compounds Nitro compounds

Unpleasant

-SH -SR -CSR -NC -NH2

Merkaptanes Thioethers Thioketones Nitriles Amines

This division should be, however, only regarded as a rough guideline, since the so-called “pleasant” odor, at higher concentrations, can easily be perceived as very “unpleasant”. Almost all solvents, such as, ketones, esters, glycols, alcohols, aromatic and aliphatic hydrocarbons, contribute to a more or less intensive smell. “Odor” is not a parameter of substance, but a summarized parameter of effects. Its determination is based on the fact that the sense of a smell can be used subjectively to evaluate certain substances (odorants). The concentration of those substances suspended in the air can be determined, which is called odorant concentration. Odor threshold or perception threshold35-38 The odor threshold is a concentration of the odorant in the air, given in mg/m³. The odor threshold corresponds to an odor unit (GE). It is the amount of odorous substance, which - distributed in 1 m³ of scent-neutral synthetic air - initiates in just 50% of the evaluators a perception of smell and in the other 50%, no response. The odor threshold is very specific to substance. It is determined in several measuring series and the results form a Gaussian distribution curve. Since this is ultimately a subjective evaluation, one should not be surprised to find more as well as, less, reliable data in literature. Examples are shown in Table 18.1.8. Odorant concentration The odorant concentration of a sample is the multiple of the odor threshold and it is determined in odor units (GE) per 1 m³ neutral air. Odor intensity39 Since the odor threshold alone is an insufficient evaluation criterion for an odorant, the increase in response with increasing odorant concentration may additionally be taken into account as a scale of reference. The increase in response is mainly material and/or mixture dependent for a given odorant concentration and is called odor intensity. [Schön, p. 68]32


Table 18.1.8. Odor thresholds [Geruchs-Immissions-Richtlinie]34 Compound

of

Odor threshold, mg/m³

Butyl acetate

0.03

Ethyl acetate

selected

1221

compounds

Compound

from

literature

Odor threshold, mg/m³

Ethanol

19.1

22

1-butanol

0.4

Benzene

16.2

Acetone

48

Toluene

7.6

Ethyl acrylate

Xylene

0.35

Dichloromethane

0.002 706

Figure 18.1.7 shows the interaction between odorant concentration and odor intensity in an artificial example.41 Odor intensity is assessed by means of a scale from 1 to 6 [VDI 3882/1]39 (Table 18.1.9). Table 18.1.9. Correlation of odor with an intensity level Odor

Figure 18.1.7. Interdependence of odorant concentration and odor intensity in an artificial example.

Intensity level

Extremely high

6

Very high

5

High

4

Moderate

3

Low

2

Very low

1

No scent

0

Hedonic odor tone An odor may be quantified by determining the odor concentration, the character of the odor (pleasant or unpleasant) is not considered. Unpleasant odors may result in deteriorating health and should therefore be avoided. By determining the hedonic odor effect, the emotional reaction initiated by an irritation to the sense of smell may be included. It should not be confused with the kind of smell (it smells like ...) or with the odor intensity (it smells “strong” to Figure 18.1.8. Evaluation of the hedonic effect of an odor “weak”). It may be determined not only for 40 sample [VDI 3882/2].

1222

Margot Scheithauer

a single odorant, but also for an odorant mixture. Odor samples of an odorant concentration above the odor threshold are ranked by evaluators according to the odorants’ position in the following response range between “extremely unpleasant” to “extremely pleasant”. In many cases of odor evaluation, there are big differences between the results among the individual evaluators, since they have a widely varied background of experience. 18.1.6 MEASUREMENT OF ODOR IN MATERIALS AND INDUSTRIAL PLANTS 18.1.6.1 Introduction Various methods are known for determining odors. One of particular interest is pupil dilatation. This is based on the fact, that pupils dilatate when a strong odor is sensed, as may be caused by concentrated ammonia. A simple sampling of odors is based on the Öko-Tex Standard 200.42 The samples are conditioned in a desiccator and their odor is subsequently judged by 6 evaluators according to a five-point scale: 1 odorless 2 low odor 3 bearable odor 4 annoying odor 5 unbearable odor Such a scale also serves the “Deutsche Gütegemeinschaft Möbel” (German Quality Association of Furniture Manufacturers) as a methodical approach for evaluation of odors of furniture, cover fabric, leather, etc. Level 3, however, is defined as “commodity typical”. It needs to be achieved in order for the respective product to have the attribute “tested for noxious substances”. In principle, odors may be determined by means of sample recognition with the help of arrays of gas sensors, so-called electronic noses. Unknown samples are compared with known samples. Hence, olfactometric investigations need to precede. Such a measuring device is applicable only in a specific case and has to be trained prior to use. Odor can become a controllable quality feature of a product. Samples of good quality can be made distinguishable from samples of bad quality. In practice, odor determination by means of the olfactometer has been widely applied and generally adopted. 18.1.6.2 Odor determination by means of the “electronic nose” The principle of odor determination in different fields of application has been discussed in detail by Moy and Collins43 and Schulz.44 Measuring principle: The substance mixture in question, which causes the odor, undergoes an overall investigation by means of a sensor array. In doing so, the device makes use of the semi-conductive properties of various metal oxides, which are on the surface of 12 sensors which react to the gaseous substances which come in contact with them. The response also depends on the temperature, humidity and flow rate of the gas. Each sensor issues its own reply signal as soon as the sample of air touches its surface. The measurement is done within about two minutes, readjustment of the sensors takes about 4 minutes. The measuring device is linked to a computer with relevant software. Each odor is, according to the twelve sensors, represented by 12 graphs showing a characteristic “profile” (“fingerprint”). These sensor-specific, time-dependent, series of


1223

electric conductivity data, which are produced by the presence of the odorous substances provide the data for the evaluation. They are processed by means of neuronal networks. The available odor samples can be shown as two or three-dimensional. Comparative or referential samples are treated in the same way. By means of sample series, they may serve the purposes of the identification of new samples. It remains an issue, that odor-relevant compounds exist in much lower concentration in most cases than the less odorous compounds, thus they are also less influential on the sensor signal. Much research effort has yet to be undertaken to apply the electronic nose in the future. 18.1.6.3 Odor determination by means of the olfactometer The method of odor determination by means of the olfactometer is based on the guidelines Richtlinien VDI 3881, pages 1 to 435-38 and VDI 3882, pages 1 and 2.39,40 Measuring principle: An odorant sample is diluted in a defined way with neutral air in an olfactometer and offered to test persons as an odor sample. The test persons are exposed to several dilution levels. Should an odor be perceived, it shall be confirmed by Figure 18.1.9. Principle of an olfactometer. 1 diluent air pressing a button. An olfactometric detera) outer air, b) synthetic air (bottle); 2 acitvated charcoal filter air from 2d to rinse by-pass, 3 glass bulb pump with mination requires at least 5, and preferably electronically controlled drive (a,b); 4 mix chamber 8 evaluators. These evaluators need to pass (with installation a); 5 evaluator’s mask; 6 sample air (if a suitability test in accordance with Guidenecessary pre-mixed); 7 glass bulb pump with electroni- line VDI 3882, p. 2.40 The principle of an cally controlled drive (a,b); 8 by-pass valve; 9 mix 45 chamber valve; 10 control terminal, a) display olfactometer is shown in Figure 18.1.9. The results are presented in odor (programme step), b) display, c) terminal keyboard; 11 printer: I - IV control line; 12 additional pump (24 l x units/m3 of neutral air and they are automatmin-1). ically displayed as the averaged data of the panel’s evaluation. 18.1.6.4 Example for odor determination for selected materials: Determination of odorant concentration in varnished furniture surfaces The odor potential of furniture is determined primarily by the applied varnishes although adhesives also play a minor role. Furniture varnishes may contain up to 80% solvents. The residual solvents remaining in the varnish thus determine the VOC and odorant concentration of furniture surfaces. A specific test method has been developed for determining the VOC and odorant concentration of coated surfaces,46-48 as shown schematically in Figure 18.1.10. Similar to formaldehyde determination,49 sample testing is done using a test chamber approach. After some atmospheric conditioning, of the varnished furniture, the samples are stored in a test chamber (typically 1 m3) under the following conditions: 23°C±0.5; 45±3% relative humidity; charged with 1 = 1 m2 of emitting sample surface per 1 m3 test chamber volume air flow rate at sample surface: between 0.1 and 0.3 m/s. The samples remain in the test chamber for

1224

Margot Scheithauer

an initial 48 hours without an air exchange in the test chamber, in order to simulate a “worst case scenario”, i.e., that, for instance, the new furniture is sealed off air-tight during transport, so that residual solvents accumulate. After 48 h of storage time without an air exchange, air samples are taken in parallel for the determination of VOC emission by means of gas chromatograph and for determination of odor by the olfactometer. Subsequently, there is an air exchange of one air change per hour, and the Figure 18.1.10. Test approach for the determination of VOC and slow-down curve of VOC and odorant concentration in coated surfaces. odorant concentration is determined. Figure 18.1.11 shows the decreasing VOC concentration depending on the storage time in the test chamber with an air exchange 1,50,46 and Figure 18.1.12 shows the decreasing odorant under the same test conditions.50,46 The curves for VOC and the odorant concentration follow a similar pattern, they are, however, not identical. While the VOC concentration in water varnish is the lowest after 11 days, the odorant concentration after this time is equally high with both DD-varnish and water varnish. This, however, does not generally apply to these types of varnish. Also DD-varnishes of different origins may differ greatly in emitting residual solvents. In single-solvent systems, it is easy to see the interdependence between VOC and odorant concentration. This does not necessarily apply to solvent mixtures. This is due to the large variations in odor thresholds and the different evaporating behavior of the various solvents. Acrylate varnishes, for example, need contain only a few µg of ethyl acrylate in order to produce odor, since odor threshold level is at 0.002 mg/m3. Thus the determination of the

Figure 18.1.11. VOC concentration depending on the storage time in the test chamber with air exchange 1.

Figure 18.1.12. Decreasing odorant concentration depending on the storage time in the test chamber with air exchange 1.


1225

VOC concentration alone is not sufficient for determining the quality of coated surfaces must also be evaluated. 18.1.6.5 Example of odor determination in industrial plants: Odor measurement in an industrial varnishing plant In a study parallel to the measurements of the VOC emission at the varnishing lines of a furniture company, as described in Section 18.1.4.3.2, 1st Example, odorant concentrations were sampled in the exhaust air from the individual chimneys by means of olfactometry.29 All conditions of measurement described in Section 18.1.4.3.2 also apply to the determination of the odorant concentration. The results are contained in Table 18.1.5 as odorant concentration in GE (odorant units)/m3 and in the mass flow in TGE/h (one thousand odorant units per hour). The evaluation of the analytical determination of individual components resulted in the main components being butyl acetate and ethyl acetate. Due to the very low odorant threshold of butyl acetate (0.03 mg/m3), as compared to other available substances, (ethyl acetate 22 mg/m3), butyl acetate may be assumed to be responsible for for the bulk of detectable odor. An immediate comparison of samples taken at the same point in time resulted in the following data: concentration of butyl acetate: 8 mg/m3 odorant concentration: 115 GE(odorant units)/m3 Adopting the value quoted in the literature for the odorous threshold of butyl acetate at 0.03 mg/m3, which equals one odorant unit, the odorant concentration estimates a butyl acetate concentration of about 4 mg/m3. This value is in the same range as the value established for butyl acetate by gas chromatography. In these comparisons one has to consider that odorant thresholds may deviate by one to two exponents to even ten, according to literature sources.51 Measurements taken in the housing area about 300 m away from the chimneys of the varnishing plants could not establish any solvent-typical components in the gas analysis. The maximum concentration value for TVOC was at 0.004 mg/m3. Also the spreading calculation, which was carried out on the basis of the determined solvent concentration, showed that there should be no significant odor annoyance in the vicinity of the emission source. The results of the spreading calculation on the basis of measured solvent concentrations are considerably more precise, since solvent concentrations may be determined more exactly than odorant concentrations. The spreading calculation was done using the software package “IMMPROG-P” by AIRINFO AG, Switzerland, and carried out according to a method of the Odor Emission Guideline of the German State of Northrhine-Westfalia/Geruchsimmissions-Richtlinie des deutschen Bundeslandes Nordrhein-Westfalen.34 REFERENCES 1 2 3

4

Council Directive 1999/13/EC of March 11, 1999 of the limitation of emissions of volatile organic compounds due to the use of organic solvents in certain activities and installations. ISO/DIS 4618-4: 1999-12-03, Paints and varnishes - Terms and definitions for coating materials - Part 4: Terms relating to raw materials. Technische Regeln für Gefahrstoffe TRGS 610 “Ersatzstoffe, Ersatzverfahren und Verwendungsbeschränkungen für stark lösemittelhaltige Vorstriche und Bodenbelagsklebstoffe”; Absatz 2.7, 04.06.1992. M. Scheithauer, I-Lack 66, 325 - 331 (1998).

1226

5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

27 28 29 30 31 32 33 34

35 36 37 38 39 40 41

Margot Scheithauer

ISO/DIS 11890-1, Draft 03/1998, Paints and varnishes - Determination of volatile organic compound (VOC) content. Part 1: Difference method (DIN ISO 11890-1). ISO/DIS 11890-2, draft 03/1998, Paints and varnishes - Determination of volatile organic compound (VOC) content - Part 2: Gas chromatographic method (DIN ISO 11890-2). Lösemittelverordnung 1995 in Österreich 872.VO im Bundesgesetzblatt für die Republik Österreich. Lackieranlagen-Verordnung 1995 in Österreich 873.VO im Bundesgesetzblatt für die Republik Österreich. ASTM D 3960-98 (1998-11-10). Standard Practice for Determining Volatile Organic Compound (VOC) Content of Paints and Related Coatings. Farbe und Lack, 105, 12/99, 111 (1999). B. Seifert, Richtwerte für die Innenraumluft; Bundesgesundheitsblatt - Gesundheitsforschung Gesundheitsschutz 3; Springer-Verlag, Berlin, 1999, pp. 270 - 278. J. Witthauer, 4. Freiberger Polymertag, Freiberg, May 27-28, 1999, conference paper A/19. A. Szalai, The use of Time: Daily activities of urban and suburban populations in twelve countries, Den Haag, Paris: Mouton, 1972. F. S. Chapin, Human activity patterns in the city, Wiley - Interscience, New York, 1974. EUWID Möbel 24, 5 (1994). F. Busato, Mod. Paint Coat., March 97, 30 - 33 (1997). H. Zeh, 4. Freiberger Polymertag, Freiberg, May 27-28, 1999, Conference paper P/2. Lackieranlagen-Verordnung in Deutschland (Varnishing Plant Regulation): Umweltbundesamt Deutschland III 3.2 - 52337/7); draft April 1, 1996. Deutsches Institut für Gütesicherung und Kennzeichnung e.V.: “RAL-UZ 38: Emissionsarme Produkte aus Holz und Holzwerkstoffen” Eigenverlag, St. Augustin, 1999, p.4. M. Große Ophoff, Holz- und Kunststoffverarbeitung, 12/96, 52-55 (1996). ASTM D 3960 - 98: Standard Practice for Determining Volatile Organic Compound (VOC) Content of Paints and Related Coatings, 1998. ISO 3251: 1993. Paints and varnishes - Determination of non-volatile matter of paints, varnishes and binders for paints and varnishes. ISO 760: 1978. Determination of water - Karl-Fischer method (general method). E DIN 55 649: 1998 - 10. Paints and varnishes - Determination of volatile organic compounds content in waterthinnable emulsion paints (In-can VOC); draft: October 1998. Richtlinie VDI 3481 / Blatt 3: Gaseous emission measurement - Determination of volatile organic compounds, especially solvent, flame ionisation detector (FID), 10.95. Richtlinie VDI 3481 / Blatt 2 E: Messen gasförmiger Emissionen (Gaseous emission measurement) Bestimmung des durch Adsorption an Kieselgel erfassbaren organisch gebundenen Kohlenstoffs in Abgasen; 11.96. Dräger, Dräger-Röhrchen Handbuch, Lübeck, 1991, 249. Richtlinie VDI 3482/Blatt 1. Gaseous air pollution measurement; gas-chromatographic determination of organic compounds, fundamentals; 02.86. M. Broege, Gutachten zu von Lackieranlagen verursachten Geruchsimmissionen, Institut für Holztechnologie Dresden, 1993. 1312 Photoacoustic Multi-gas Monitor, Product Data from INNOVA, Air Tech Instruments A/S, Denmark, 04/97. R. Kusian, M. Henkel, Forschungsbericht des Institutes für Holztechnologie Dresden, Germany (1997): Untersuchung der Emissionsverhältnisse bei der Oberflächenbehandlung von Holz und Holzwerkstoffen. M. Schön, R. Hübner, Geruch-Messung und Beseitigung, Vogel-Buchverlag, Würzburg, 1996. A. L. Lehninger, Biochemie,Weinheim-New York, Verlag Chemie, 1979, p. 19. Feststellung und Beurteilung von Geruchsimmissionen (Geruchsimmissionsrichtlinie, GJR); Der Minister für Umwelt, Raumordnung und Landwirtschaft des Landes Nordrhein-Westfalen, Stand 15.02.1993, Anhang B.Richtlinie VDI 3881. Blatt 1: Olfactometry; odor threshold determination; fundamentals; 05 / 1986. Blatt 2: Olfactometry, odor threshold determination; sampling; 01/1987. Blatt 3: Olfactometry; odor threshold determination; olfactometers with gas jet dilution; 11/1986. Blatt 4: Olfaktometrie; Geruchsschwellenbestimmung; Anwendungsvorschriften und Verfahrenskenngrößen; 12/86 Richtlinie VDI 3882. Blatt 1: Olfactometry; determination of odor intensity; October 1992. Blatt 2: Olfactometry - Determination of hedonic odor tone, September 1994. M. Paduch, VDI-Berichte 1059, “Aktuelle Aufgaben der Messtechnik in der Luftreinhaltung” zum Kolloquium Heidelberg, Düsseldorf, Juni 2-4, 1993, pp. 593-607.

18.2 Prediction of organic solvents emission

42

43 44 45 46 47 48 49 50 51

1227

Öko-Tex Standard 200: Prüfverfahren für die Vergabe der Berechtigung zur Kennzeichnung von Teppichböden mit “Schadstoff geprüft nach Öko-Tex Standard 100" (Österreichisches Forschungsinstitut) 1992. L. Moy, M. Collins, LaborPraxis - Journal für Labor und Analytik, 20/5, 14-18 (1996). H. Schulz, 4. Freiberger Polymertag, Freiberg, May 27-28, 1999, conference paper C 4/5. Ströhlein, Labor-, Mess- und Umwelttechnik, Kaarst, Germany, Product Data p. 25, principle of an olfactometer. M. Scheithauer, K. Aehlig, M. Broege, Holz- und Kunststoffverarbeitung, 1/96, 58-61 (1996). M. Scheithauer, K. Aehlig, Konferenz im ITD, Poznan (Poland), 1995. K. Aehlig, M. Scheithauer, M. Broege, Holz, 5, 26-32; (1996). prENV 717-1: 1998: Holzwerkstoffe Bestimmung der Formaldehydabgabe Teil 1: Formaldehydabgabe nach der Prüfkammer-Methode (1998). M. Broege, K. Aehlig, 4. Freiberger Polymertag, Freiberg, Mai 27-28, 1999, conference paper R 4/5. G. Scharfenberger, Chemie in Labor und Biotechnik, 42, 498-502 (1991).

18.2 PREDICTION OF ORGANIC SOLVENTS EMISSION DURING TECHNOLOGICAL PROCESSES Krzysztof M. Benczek, Joanna Kurpiewska Central Institute for Labor Protection, Warsaw, Poland

18.2.1 INTRODUCTION The concentration of toxic substances in air during technological process is very important factor for occupational safety. Typical examples of processes, which have the potential to harm workers, are metal degreasing, painting, and wood impregnation. If metal processing involves several steps some of which may be done in more than one manufacturing facility, the semi-processed metal parts must be protected during transportation and storage. Such protective coatings of grease and rust preventatives must be removed in degreasing operation. For many processes (e.g., painting, galvanic metal deposition), clean surface is an important requirement. The cleaning process may be done in automated and enclosed equipment or it may be done manually in the open. The degreasing agent may be an organic solvent, a solvent blend, or a water solution, usually alkaline in nature. Depending on the process used the operation may pose no risk to the worker or be a serious occupational hazard. Similarly environmental emissions may be negligible or of serious concern. We present a method of evaluating the quantities of emissions from such processes which involve solvents. The method may be applied to such diverse operations as painting, wood preservation, impregnation of porous materials, gluing, cleaning, filing open tanks, general solvent handling operations, and many others. We have selected metal degreasing as a representative example to demonstrate how the method may be applied. 18.2.2 METHODS OF DEGREASING Six methods can be identified which differ in the degreasing agent used:

1228

Krzysztof M. Benczek, Joanna Kurpiewska

• degreasing in liquid organic solvents, such as naphtha, petroleum, chlorinated aliphatic hydrocarbons, etc.

• degreasing in hot vapors of halogenated solvents such as, trichloroethylene, perchloroethylene, fluorochloroethane and so on

• degreasing in alkaline water solutions of hydroxides, phosphates, surfactants, emulsifiers, common inhibitors, etc.

• degreasing in an emulsion of organic solvents in water • degreasing in water steam • supersonic degreasing in stabilized chlorinated hydrocarbons. Degreasing process can be conducted automatically in different ways by: dipping spraying using high pressure supersonic pulsating washing Process can be conducted in open or enclosed equipment. Manual degreasing is still very popular but it is very time and labor consuming, expensive, and large amounts of solvents are lost. 18.2.3 SOLVENTS In metal degreasing these solvents are most frequently used: • naphtha solvent, • naphtha anti-corrosive • 1-butanol, • 1,1,2-trichloroethylene, • 1,1,1- trichloroethane, • extraction naphtha, • petroleum(mineral) spirits. 18.2.4 IDENTIFICATION OF THE EMITTED COMPOUNDS In many cases, solvent mixtures are used. Their composition must be identified. Rodofos is one example of such solvent used in Poland. Its composition was determined by gas chromatography. Analyses were performed using a Hewlett-Packard gas chromatograph model 5890 coupled with computerized mass spectrometer instrument, model 5970. Capillary column 50 m x 0.32 mm i.d., d.f.= 0.52 µm FFAP and helium as a carrier gas was used at temperature of 40oC. Ions from 20 to 400 amu were counted, delay time was 3.5 min. Samples were collected by drawing a known volume of air through a bubbler containing 1 ml of carbon disulfide. Volume of the injected sample was 1-5 µl. Chromatograms are presented in Figure 18.2.1. 18.2.5 EMISSION OF ORGANIC SOLVENTS DURING TECHNOLOGICAL PROCESSES The concentration of substances emitted to the air during the degreasing processes reached the steady-state constant value:

• • • • •

C = E/q

[18.2.1]


1229

Figure 18.2.1. Chromatogram of substances emitted during degreasing. Oven temperature 40oC. 1 - benzene, 2 2,2-dichloromethylpropane, 3 - butyl ester of formic acid, 4 - methylbenzene, 5 - water, 6 - 1-butanol, 7, 8, 10, 11, 12, 13, 14 - derivatives of methyl ethyl benzene, 9 - propyl benzene. where: C q E

concentration, mg/m3 sum of ventilation flow rates, m3/min emission, mg/min.

Thus, E = C×q

[18.2.2]

The value C can be measured, but a value for q is difficult to estimate, because it includes mechanical and gravitational ventilation (central air conditioning, influence of open doors, windows, fans and natural ventilation). When the process of degreasing ends, the decay of concentration can be described by: Ct = Ce × exp(-t × q/V)

[18.2.3]

where: Ce t V

concentration of the emission at the termination of the process, mg/m3, duration of process, min, room volume, m3.

After a transformation: q = (lnCe - lnCt) × V/t

[18.2.4]

The quantity emitted to a room of known volume depends only on the changes of concentration at time (t).

1230


18.2.6 VERIFICATION OF THE METHOD Emission measuring setup is given in Figure 18.2.2. Cyclohexane was pumped (with E=6.123 mg/min) into glass bottle (V=44.5 l) equipped with a fan. The air in the bottle was analyzed by gas chromatography using a flame ionization detector - OVA 118. After 20 min., when the concentration reached 2.250 mg/l, the pump was stopped and the decay of concentration with time was measured. Results are given in Tables 18.2.1 and 18.2.2 (value E from the equation [18.2.2], q from the equation [18.2.4]). The standard deviation of the sample was 0.0106, of the method - 0.0102 and bias was 0.08 mg/min (1.3%). Tests were then performed in the macro-scale - emission of cyclohexane in a room, when the emission ended when the process stopped. The estimate is precise when the room volFigure 18.2.2. Emission measur- ume and ventilation flow rate are known. ing setup. The method was verified in the industrial environment in automotive plant. Here, the hazardous substances continue to be emitted to the air after the process had stopped. The concentration measured near the outTable 18.2.1. Emission of cyclohexane

Table 18.2.2. Ventilation flow rate

t, min

Ct, mg/l

E, mg/min

t, min

Ct, mg/l

Q, l/min

1

0.138

6.210

1

2.200

1.000

2

0.273

6.212

2

2.215

1.011

3

0.405

6.213

3

2.100

1.024

4

0.533

6.201

4

2.055

1.008

5

0.659

6.201

5

2.010

1.004

6

0.780

6.184

6

1.965

0.993

8

1.020

6.200

8

1.880

0.999

10

1.250

6.212

10

1.800

0.992

12

1.465

6.200

12

1.720

0.996

15

1.770

6.188

15

1.605

1.002

20

2.250

6.218

20

1.435

E ave. - 6.203, σ n= 0.0102, σn-1= 0.0106

1.000

q ave. = 1.0026, σn = 8.7103×10 , σn-1 = 9.1354×10-3 -3

let of exhaust was 620 mg/m3, and because the exhaust flow rate was known to be 50 m3/min, the emission was precisely estimated according equation [18.2.2] as 31 g/min or 312.48 kg/week. This value was comparable with the average solvent loss - 326.22 kg/week.


1231

18.2.7 RELATIONSHIPS BETWEEN EMISSION AND TECHNOLOGICAL PARAMETERS Emission of solvents depends on the evaporation rate of the solvent in the process. The evaporation rate from the surface depends on the concentration in the layer on the surface and the coefficient of mass transfer on the air-side. This relation is approximately true for degreasing operations using both liquid and vapors of organic solvents. The concentration of solvent in a saturated vapor layer depends on temperature and vapor pressure. The coefficient of mass transfer on the air-side depends on the air velocity in the layer on the surface and Schmidt's number (includes dynamic vapor viscosity, vapor density, and diffusion coefficient). Emissions are measured in mass unit per unit of time and the amount depends on surface area and the rate of evaporation, which, in turn, depends on temperature, air velocity over the surface of solvent and the mass of solvent carried out on the wetted parts which have been degreased. 18.2.7.1 Laboratory test stand A thin-metal, flat dish 6 cm in diameter was filled with solvent up to 2-3 mm from the upper edge and placed on a laboratory balance (Figure 18.2.3). The amount of evaporated solvent was measured with (precision 0.1 mg) as the difference between the mass of the dish and solvent at the start of the test and the mass of the dish with solvent after pre-determined period of time. The test was repeated under different conditions of temperature and air velocity near the surface of the solvent. The results were reported as the evaporated mass per 1 minute. 18.2.7.2 The influence of temperature on emission Temperature was measured with a mercury thermometer with a range from 0 to 30oC. Air velocity was 0.3 to 0.4 m/s in this temperature region. The relationship is linear and the equations expressing emission [g/m3/h] relative to temperature [oC] for different solvents are: trichloroethylene E= 63.6 × t + 699.6 naphtha solvent E= 9.3 × t + 4.9 naphtha anti-corrosive E= 0.55 × t + 5.65 1-butanol E= 6.29 × t + 16.9 1,1,1- trichloroethane E= 94.7 × t + 805.6 extraction naphtha E= 49.5 × t + 1147 petroleum(mineral) spirits E= 11.9 × t + 76.7 wood preservatives E= 2.0 × t + 42

Figure 18.2.3. Test laboratory stand.

1232


18.2.7.3 The influence of air velocity on emission Air velocity was measured with a TSI air velocity meter in range 0.2- 1.5 m/s. Temperature was 20oC. The relationship has linear character and the equations expressing emissions [g/m3/h] relative to air velocity [m/s] for different solvents are: trichloroethylene E= 657 × v + 1766 naphtha solvent E= 109 × v + 159 naphtha anti-corrosive E= 9.1 × v + 14.53 1-butanol E= 72 × v + 121 1,1,1- trichloroethane E= 1403 × v + 2120 extraction naphtha E= 632 × v + 830 petroleum(mineral) spirits E= 160 × v + 242 wood preservatives E = 516 × v 18.2.7.4 The relationship between the mass of solvent on wet parts and emissions The amount of solvent carried out on a degreased part depends on its surface, shape and roughness and on the viscosity and density of the solvent. The amount of solvent carried out on a degreased surface was measured by gravimetric method. Tests show that the amount of solvent retained on 1 m2 of flat surface is: trichloroethylene 34.7 g naphtha solvent 26.0 g naphtha anti-corrosive 34.5 g 1-butanol 31.3 g 1,1,1- trichloroethane 38.5 g extraction naphtha 19.0 g petroleum(mineral) spirits 19.2 g 18.2.8 EMISSION OF SOLVENTS Combining these factors, the final equations for the evaluated solvents are: trichloroethylene naphtha solvent naphtha anti-corrosive 1-butanol 1,1,1- trichloroethane extraction naphtha petroleum(mineral) spirits wood preservatives

E= (657v + 63.6t + 699.6) × Pr + 34.7P E= (109v + 9.3t + 4.9) × Pr + 26P E= (9.1v + 0.55t + 5.65) × Pr + 34.5P E= (72v + 6.3t + 17) × Pr + 31.3P E= (1403v + 94.7t + 805.6) × Pr + 38.5P E= (632v + 49.5t + 1147) × Pr + 19P E= (160v + 12t + 76.7) × Pr + 19.2P E= (516v + 2t + 42) × Pr

[18.2.5] [18.2.6] [18.2.7] [18.2.8] [18.2.9] [18.2.10] [18.2.11] [18.2.12]

where: E v t Pr P

emission of the solvent, g/h air velocity near the surface of solvent (in still air it is 0.3 m/s) temperature, oC the surface of solvent, m2 the surface of elements degreased in one hour, m2/h.

18.2.9 VERIFICATION IN INDUSTRIAL CONDITIONS Research was conducted in an industrial hall where an ALDI machine was used for degreasing with trichloroethylene vapor. The volume of the solvent was 1000 dm3, the surface area was 0.825 m2 (110 cm by 75 cm). A metal basket with degreased parts was intro-


1233

duced into the degreasing compartment and vapors of trichloroethylene condensed on surface of the parts. After 30 s, the basket was removed from the machine. The average temperature at the surface was 24oC. There was an exhaust in the hall with a flow rate of 50 m3/min. The solvent loss was 30 l/day and it was independent of the number of degreased elements. The solvent loss of 30 l/day equals 1.25 l/h or 1.825 kg/h with density of trichloroethylene equal 1.46 g/cm3. The average concentration of trichloroethylene measured near the exhaust by portable IR spectrophotometer-Miran was 680 mg/m3. Calculation (equation [18.2.2]) with q=50 m3/min gives emission of 34 g/min or 2040 g/h. According to equation [18.2.5] for trichloroethylene: E= (657v + 63.6t + 699.6) × Pr + 34.7P v=0.3 m/s (still air near the surface) t=24oC Pr=0.825 m2 In vapor degreasing, the coefficient of mass carried out on wet details ready to ship P=0. Then E= (657 × 0.3 + 63.6 × 24 + 699.6) × 0,825 = 1999 g/h The measured value of 2040 g/h and the calculated value of 1999 g/h are in good agreement, meaning that the equations can be successfully applied to predict the organic solvents emission during process of automatic degreasing. Process of manual degreasing was verified for washing motor parts in a metal dish 0.72 m2 filled with 500 l of extraction naphtha. The surface of parts was 0.227 m2. This process took place in an open garage at temperature 14oC. Air velocity in vicinity of parts was 0.48 m/s. The residual solvent volume after degreasing was 4.58 l. The process lasted 7 minutes. Emission estimated according the loss of the solvent during the process was 420 ml. Taking into account the weight density of naphtha - 0.72 g/cm3, emission was equal 302.4 g/7 min or 2592 g/h. Applying the equation [18.2.10] for extraction naphtha E= (632v + 49.5t + 1147) × Pr + 19P and: v = 0.48 m/s t = 14oC Pr = 0.72 + 0.227 = 0.947 m2, surface of the dish and details P = 0.947 m2/7min [8.12 m2/h], solvent carried out on the details and on the surface of the dish after pouring out extraction naphtha Thus: E= (632 × 0.48 + 49.5 × 14 + 1147) × 0.947 + 19 × 8.12 = 2184 g/h Again measured value of 2592 g/h and the calculated value of 2184 g/h were in good agreement thus the equation can be applied for predicting the organic solvents emission during the processes of manual degreasing.

1234


18.3 INDOOR AIR POLLUTION BY SOLVENTS CONTAINED IN PAINTS AND VARNISHES Tilman Hahn, Konrad Botzenhart, Fritz Schweinsberg Institut fòr Allgemeine Hygiene und Umwelthygiene Universit¬t Thbingen, Thbingen, Germany


18.3.1 COMPOSITION - SOLVENTS IN PAINTS AND VARNISHES. THEORETICAL ASPECTS Solvents are normally defined as fluids having a maximum b.p. of 250oC, which are able to dissolve other components of paints, especially binders. They evaporate under drying conditions when paint films are formed.1,2,3,4 Solvents must not react with the painted or varnished product. The composition of paints and varnishes is determined by application requirements, drying temperature, and drying time.2,5 Depending on the properties of paints and varnishes, different mixtures of solvents are added. Functions and properties of solvents in paints and varnishes:2,5 • Dissolve several components, especially binders • Influence and control paint viscosity • Wet pigments (influence on solubility, hydrogen bonding by solvents, prevent the separation of pigment) • Influence and control flow properties (e.g., butyl acetate, butanol, glycol ethers) • Influence skin formation.6 The aim is to produce a homogeneous cure when the paint or varnish film hardens without the formation of a stable surface film during the drying period. The correct composition of the solvent will avoid trapping solvents under the surface film. • Influence drying process. Acceleration by low boilers, production of a flawless surface by medium and high boilers (chemical and physical drying processes) • Influence surface tension (e.g., increase by rapid evaporation of solvents) • Influence mechanical properties of paints (e.g., adhesion properties) • Influence blushing or blooming of paints by preventing the absorption of condensed water by various solvents, e.g., ethanol and glycol ether • Influence gloss (e.g., improvement with high boilers) • Prevent defects particularly in varnish coating (e.g., background wetting) • Influence electrostatic properties (spray painting) • Influence defined surface properties (structural change) • Influence durability of paints and varnishes • Influence product suitability, e.g., spraying and dipping lacquers which need to dry at room temperature In addition to their effect on performance and properties solvents interact with other components in paints and varnishes in significant ways. Interactions between binders and solvents in paints and varnishes are very important.2,5 With the aid of solubility parameters solvents or mixtures of solvents which produce

18.3 Indoor air pollution by solvents

1235

the required properties may be selected. The influence extends to dissolving of binders, reduction of paint viscosity, pigment wetting, etc. Optimized dissolving of binders can be achieved by selecting the appropriate solvent mixtures, in which the density approaches that of the binder solubility range. Nevertheless the selection of an optimized solvent mixtures is difficult because there are conflicting requirements and outcomes.2,5 On one hand, the chemical nature of the solvents should be similar to those of the binder to improve the flow but on the other hand, the solubility and hydrogen-bonding abilities of the solvents should be at the edge of the binder solubility range, because this results in rapid drying with low retention of solvents. 18.3.2 OCCURRENCE OF SOLVENTS IN PAINTS AND VARNISHES 18.3.2.1 Solvents in products Classification Authorized commissions in Germany describe various groups of solvents in paints and varnishes: The Association of Varnish Industry VdL (Verband der Lackindustrie) classifies solvents generally used in paints and varnishes in the following groups:4 Aliphatics, alcohols, aromatic hydrocarbons, esters, glycols, glycol ethers, ketones, terpenes, gasoline, water. The commission of Hygiene, Health and Environmental Protection (Hygiene, Gesundheit und Umweltschutz) which belongs to the German Standards Commission DIN (Deutsches Institut für Normung e.V.) also describes solvents which may be present in paints as follows:10 • Aliphatics (white spirit, turpentine substitutes, cyclohexane) • Aromatic hydrocarbons (toluene, xylenes, naphtha, styrene) • Alcohols (methyl alcohol, ethyl alcohol, propyl alcohol, butyl alcohol, glycols) • Ketones (acetone, methyl ethyl ketone, methyl isobutyl ketone, cyclohexanone) • Esters (methyl acetate, ethyl acetate, butyl acetate) • Others: methylpyrrolidone, oils of turpentine Chlorinated hydrocarbons are not contained in the above list. Benzene is not included as it is obsolete, although it is sometimes found in some petroleum solvent-based paints. Shortcomings The description of paints and varnishes is usually neither complete nor reliable: • Instructions and comments from manufacturers and suppliers differ in quantity and quality. Usually manufacturers or suppliers are not committed to indicate the exact details of their product’s composition. They often omit information (e.g., information concerning product components, additives or by-products). • Lack of standards. In Europe there are some mandatory standards concerning products used in construction (e.g., asbestos, formaldehyde in particle board, organic solvents in certain activities and installations are prohibited). • Solvents or other organic compounds may be present, in low concentrations, in some products (even in water-based paints), e.g., as additives or by-products although there is no reference to them. • Additives (e.g., low molecular compounds such as esters or glycol ethers with high boiling points) serve as aids for the formation of films, in repelling water, in assisting plasticization or for other functions. In solvent-based paints, additives or diluting agents are often intentionally mixed in.

1236


Table 18.3.1. Systematology of paint and varnish products (after reference 10) 1. Paint coatings 1.1 Coloring agents 1.1.1 soluble pigments 1.1.1.1 natural pigments 1.1.1.2 synthetic pigments 1.1.2 insoluble pigments 1.1.2.1 inorganic pigments 1.1.2.2 organic pigments 1.1.2.2.1 animal and vegetable pigments 1.1.2.2.2 synthetic pigments 1.2 Binders 1.2.1 water dilutable binders 1.2.1.1 slaked lime (lime colors) 1.2.1.2 standard cement (cement colors) 1.2.1.3 sodium silicate (colors of 1 or 2 components) 1.2.1.4 vegetable glues (limewash) 1.2.1.5 casein (lime-casein products, alkali-lime products) 1.2.1.6 dispersions 1.2.1.6.1 natural resin emulsion paints 1.2.1.6.2 plastomer emulsion paints (PVAC (homopolymers, copolymers), PVP, polyacrylates (PMMA, styrol-acetate)) 1.2.1.6.3 water emulsifiable varnish systems (aqueous acrylate systems, aqueous polyurethane systems) 1.2.2 solvent dilutable binders 1.2.2.1 oil paints 1.2.2.2 varnishes 1.2.2.2.1 products drying by air oxidation (nitrocellulose varnishes, aldehyde resin varnishes, oil varnishes) 1.2.2.2.2 physical drying products (polymer resin varnishes, polyvinyl chloride varnishes, polyvinyl acetate varnishes, polyacrylate varnishes, chlorinated rubber varnishes) 1.2.2.2.3 chemical curing products (phenolic varnishes, aminoplast varnishes, melamine resin varnishes) 1.2.2.2.4 acid-curing varnishes 1.2.2.2.5 epoxy resin varnishes 1.2.2.2.6 polyurethane varnishes 1.2.2.2.7 unsaturated polyester varnishes

• Residuals or by-products can result from various processes (e.g., residual monomers in a polymerization processes).

• Substances which can be classified as solvents are not always exactly defined, e.g., • •

substances in paints with maximum boiling point above 250oC. Plasticizers and solvents cannot often be strictly separated (e.g., various SVOCs). Analytical problems and difficulties in assessment (e.g., mixtures of substances, very low concentrations). Even new products are often not well documented (e.g., concerning composition, see above).


1237

18.3.2.2 Paints and varnishes Definitions and systematology of paints and varnishes are given by various authorized commissions, e.g., European Committee for Standardization (CEN)1 and DIN-Commissions9 (see also Table 18.3.1). Various databases list the paint and varnish products and their composition generally applied in Europe, see Table 18.3.2). Table 18.3.2. Types of paints and varnishes mainly used in Europe - product groups, important compounds, applications (after references 2,5,11,23,25) Product groups

Compounds

Application

Silicate paints/prod- potash-waterglass (binder), inorganic/organic pigwall paints, outside house ucts, emulsion paints ments, mineral fillers, synthetic resins (dispersions, paints DIN 18363 (M-SK01) stabilizers), water (dispersing agent) water- and weatherproof Silicate paints/prod- potassium silicate (binder, fixing solution), inorganic painting, interior coating ucts, (M-SK02) pigments, adjuvants (resistant to chemicals) Emulsion paints (M-DF02)

synthetic resins (dispersions), mineral fillers, inorganic/organic pigments, water (dispersing agent), ad- outside house paints, interior ditives (film-forming agents < 3 %: e.g., glycol-ethers, coating (wall paints) esters, glycols, hydrocarbons), formaldehyde < 0,1 %

synthetic resins (dispersions), mineral fillers, inorEmulsion paints, ganic/organic pigments, water (dispersing agent), adinterior coating (wall paint) solvent-free (M-DF01) ditives (film-forming agents: no solvents), formaldehyde < 0,1 % Emulsion paints, outside house paint, water dilutable

emulsions of plastic materials (acrylate, vinyl), pigments, water (dispersing agent), organic solvents: < 0,1-4 %, (glycols, glycol-ethers, mineral varnish)

Emulsion paints, varnishes (M-LW01)

synthetic resins, alkyd resins, copolymerizates, polyurethane resins, inorganic/organic pigments, mineral fillers, additives, water (dispersing agents), organic various applications solvents: 5-10 %, glycols, glycol-ethers, esters, mineral varnish

Aldehyde resin varnishes, aromatic compounds (M-LL03)

aldehyde resins (binders), inorganic/organic pigments, fillers, 30-55 % solvents: mineral varnish (mixture of covering varnishes, primers hydrocarbons), other solvents (< 10 % esters, ethers, alcohols)

Aldehyde resin varnishes, low levels of aromatic compounds (M-LL02)


Aldehyde resin no aromatic substances (M-LL01)


exterior coating

1238

Product groups


Compounds

Application

Polymer resin paints, copolymers, inorganic/organic pigments, fillers, 35-50 outside paints (mineral high levels of aromatic % solvents: mineral varnish (mixtures of hydrocarbackground) compounds (M-PL03) bons), other solvents (< 10 %, esters, ethers, alcohols) Polymer resin paints, copolymers, inorganic/organic pigments, fillers, 35-50 outside paints (mineral backlow levels of aromatic % solvents: mineral varnish (mixtures of hydrocarground) compounds (M-PL02) bons), other solvents (< 10 %, esters, ethers, alcohols) Polymer resin paints, no aromatic compounds (M-PL01)

copolymers, inorganic/organic pigments, fillers, 35-50 outside paints (mineral back% solvents: mineral varnish (mixtures of hydrocarground) bons), other solvents (< 10 %, esters, ethers, alcohols)

Polymer resin paints, copolymers, inorganic/organic pigments, fillers, 35-50 outside paints (mineral backdilutable with solvents % solvents: mineral varnish (mixtures of hydrocarground) (M-PL04) bons), other solvents (< 10 %, esters, ethers, alcohols)

Natural resin paints

natural resins (e.g., shellac) or chemical modified natural resins (e.g., colophonium derivates), additions (e.g., methyl cellulose, natural latex, casein), inorganic, organic pigments (mainly natural origin), min- various applications eral fillers, additives (organic solvents: alcohols, terpenes, oil of turpentine, limonenes), essential oils (eucalyptus oil, oil of rosemary, oil of bergamot)

natural resins (e.g., shellac) or chemical modified natural resins (e.g., colophonium derivates), additions Natural resin paints, (e.g., methyl cellulose, natural latex, casein), inor- various applications solvent-free ganic, organic pigments (mainly natural origin), mineral fillers, additives (see above): < 1 % Oil paints, terpene products (M-LL04)

oils (linseed oil, wood oil, soya oil), natural resins and modified natural resins, mineral pigments, wetting covering varnishes, primers agent, flow improver, solvents: oils of turpentine, isoaliphatics, terpenes (citrus, orange)

oils (linseed oil, wood oil, soya oil), natural resins and Oil paints, terpene-free modified natural resins, mineral pigments, wetting covering varnishes, primers (M-LL05) agents, flow improvers, solvents: isoaliphatics (dearomatized) Oil paints solvent-free standard oils, calcium carbonate, pigments, siccatives, exterior coating (paneling) No. 665 lemon oil water-soluble Clear lacquers/glazing composition (wood), low levels of aromatic compounds (M-KH03)

alkyd resins, nitro resins, polymer resins, pigments, fillers, 40-50 % solvents: mineral varnish (mixtures of hydrocarbons), other solvents (< 10 %, esters, ethers, alcohols)

Lead chromate enamels, aromatic compounds

synthetic resins (e.g., aldehyde resins, PVCpolymerizates) inorganic/ organic pigments (lead priming coat (steel, alumichromate up to 20 %), fillers, 30-55 % solvents (mixnum, zinc) tures of mineral varnish, glycol-ethers, aromatic compounds)

interior coating (clear lacquers), exterior and interior coatings (glazing compositions, wood)


Product groups Silicone resin products, water dilutable (M-SF01)

1239

Compounds

Application

emulsions of silicone resins mineral fillers, inorganic/organic pigments, water (dispersing agent), additives (film forming agents < 3%)

Solvent composition is an important aspect in classifying paints and varnishes.5 Main groups are: • Solvent products. These products contain solvents of various mixtures, types and concentrations depending on the properties desired (e.g., application method, surface film or skin formation, see above 18.3.1, 18.3.2.1, Table 18.3.1 and 18.3.2). Solvents are normally the main components of these products (for example up to 80 % in nitrocellulose varnishes, low solids). • Solvent reduced products. These contain solvents in lower concentrations compared to conventional products and hence have a higher content of solids.1,2 The main groups of solvent reduced paints are medium solid contents (solids 55-65 %) and high solid contents (solids 60-80 %). • Waterborne coatings. This group contains deionized water as a dispersing agent. Normally cosolvents are added (up to 25 %). The term “waterborne coatings” is mainly applied to industrial coating materials, which differ from silicate colors, wood preservative varnishes and emulsion paints. • Solvent-free products. The products of this group are produced and applied without organic solvents: powder coatings, radiation curing systems, solvent-free water coating (without cosolvents). The use of paints and varnishes containing high solvent concentrations is becoming less common, solvent-reduced products, waterborne coatings and solvent-free products are applied more often. Whereas environmental and health-related concerns call for the reduction of solvents in paints and varnish products (see 18.3.3.2 and 18.3.4), qualitative aspects still demand the use of solvents in some fields of application.12 Solvent-reduced products should achieve the same qualitative properties as solvent-containing products (e.g., application properties, periods of guarantee, limited costs, loading capacities, surface properties). The development of solvent-reduced or solvent-free varnish products with high quality (e.g., durable, good application properties) and limited costs must be encouraged if not mandated. In some fields of application (e.g., waterborne coatings, high solids in varnish coatings of vehicles) a lot of developmental work has already been done. Additionally, other components in varnishes apart from solvents or modifications of application techniques can improve the properties of solvent-reduced paint and varnish products. Nevertheless, a wide rage of quality exists in paints and varnish products which are offered commercially and, in some cases (e.g., concerning film forming processes, processibility, corrosion protection, purification, special applications or wood preservation), solvent-based products are still preferred. In the field of wood preservation especially, solvent-based products (alkyd resins) are used because of some technical advantages (e.g., more solid surfaces) but alternative high solid systems are available.12

1240


18.3.3 EMISSION OF SOLVENTS 18.3.3.1 Emission Solvents are usually the most significant emission products coming from building materials and interior furnishings.2,5,6,13 All painted products are potential sources of emission. Even the so-called “bio” paints or natural paints emit various substances,5 e.g., mineral varnishes, natural oils and even synthetic terpene-like compounds. Depending on the products and the components which make them up, the various parameters listed below can determine the emission and behavior of solvents in ambient air from paints and varnishes:2,5,6 • Film formation. During the film formation stage solvents are emitted. The emission rate is directly proportional to the VOC concentrations in the product and inversely proportional to the film thickness (first order of kinetics). When the film has completely formed, the emission is controlled by diffusion processes, the emission rate is now inversely proportional to the square of the film thickness. • Application of the paints and varnishes, methods of application of the paint or varnish, e.g. speed of application of the paint. • Characteristics of solvents in paints and varnishes, e.g., volatility (boiling point), dynamic characteristics of evaporation and concentrations. Substances having a low boiling point evaporate fast, mostly during application and cause a rapid skin formation. Thus the risk of exposure is mainly with the painters. Medium boilers allow the surface to remain open for a while (evaporation of volatile products). The evaporation of substances with a high boiling point is slow, taking several weeks or months after application, resulting in exposure to the building occupants. • Characteristics of other compounds in paints and varnishes (e.g., relationship of binders to solvents, possible reactions between solvents and other compounds). • Characteristics of surfaces which have been painted (e.g., area, structure of surface). • Characteristics of emission processes, e.g., type of emissions (e.g., diffusion), dynamics of emissions (constant of evaporation), interrelations (e.g., diffusion and back diffusion). The quantitative assessment of emission processes can be described with various models. The usefulness of these models differs. Some models describe these processes very well, as proven by various experiments or measurements (e.g., test chambers). Basic equations which describe emission processes are shown in Table 18.3.3. The emission processes of solvents from paints and varnishes can be divided into two phases:2,5,6 1. Emissions during application of paints. This deals with complex interrelations dependent on various parameters. 2. Emissions after application process. Here the course is governed by complex emission processes dependent on various parameters (e.g., film formation, surface area). Most solvent products, especially organic solvents and some additives, emitted from paints and varnishes are VOCs. The largest components of VOCs are solvents, e.g., aliphatic and aromatic hydrocarbons, alcohols, amines, acids, aldehydes, esters, ketones, terpenes. The definition of the term VOC varies, a standard definition is published by CEN (European Committee for Standardization):1 VOCs are any organic liquids and/or solids that evaporate spontaneously at the prevailing temperature and pressure of the atmosphere.


1241

VOCC (volatile organic compound content) is defined as follows:1 Mass of the volatile organic compounds in a coating material, as determined under specified conditions. Table 18.3.3. Example for basic calculations of VOC-emissions during application of emulsion paints (after references 2,5,6,13) 1. At the beginning of the application process (t=0) the mass of VOC changes positively (vSTdscw) , on ht e other hand VOC evaporates (first order of kinetics). dmw/dt = vSTdscw - klmw/ds [18.3.1] 2. In the ambient air the mass of VOC increases because of the evaporation out of the wall and decreases according to the ventilation rates. dmL/dt = k1mw/ds - k2mL [18.3.2] 3. If the connected differential equations are solved and integrated (from t=0 until the end of application t=A/vST), the following equations are received: mw(t) = vSTds2cw/k1(1-exp(-klt/ds)) [18.3.3] and [18.3.4] mL(t) = ((1-exp(Bt)/B - (1-exp(k2t)/k2)vSTdscwexp(-k2t)) with B = k2 - k1/d2 [18.3.5] 4. After finishing application only evaporation is relevant (equation [18.3.1] is simplified): dmw/dt = - k1mW/ds [18.3.6] 5. The course of VOC in the ambient air does not change (equation [18.3.2] corresponds to equation [18.3.7]): dmL/dt = k1mw/ds - k2mL [18.3.7] 6. The solution of these differential equations describes the quantities of VOC in the wall (equation [18.3.8] and the course of VOC in the ambient air [18.3.9]: mw(t) = mw,AEexp(-k1(t-tAE)/ds) [18.3.8] mL(t) = ((k1mw,AE/ds) (exp(B(t- tAE))-1)/B + mL, AE) exp(-k2(t-tAE)) [18.3.9] with B = k2 - k1/ds [18.3.5] where: A area of the wall a coating thickness B fraction of binder cW VOC-concentration in the wall cL VOC-concentration in the ambient air D density ds thickness of the layer of the paint application (=a/D) k1 constant of evaporation k2 ventilation rate of the indoor air mL mass of VOC in the indoor air mL,AE mL at the end of the application mw mass of VOC in the wall mW,AE mW at the end of the application RM VOC-content in the dispersion t time V volume of the indoor air vST spreading velocity

In VOCs, especially these emitted from coating materials, the evaporation temperature is specified by European regulations: b.p. max. 250oC (according to DIN ISO 11890-1,-2 or 96/13/EC)14 or b.p. max. 260oC (according to VDI Guidelines 4300-6).15 In

1242


some US governmental legislation VOCs are defined solely as those compounds that are photochemically active in the atmosphere (ASTM D 3960).1 Regulations and assessment of VOCs are under discussion in Germany, e.g., discussion of various threshold limit values of TVOC (total volatile organic compounds): < 200 µg/m3 according studies of Molhave16 and 300 µg/m3 according to BGA.17 The percentage of VOC emissions caused by solvents has increased recently mainly because of the reduction of VOC coming from vehicular traffic, e.g., in Germany the estimated anthropogenic VOC emissions (without methane) caused by solvents in 1994 was about 51% compared to 37% in 1988.12 The solvent VOC emissions are mainly connected with coating materials which are responsible for more than 50% of solvent-based VOC emissions in Germany in 1995 (about 38% caused by varnishes, 4 % by building materials, 19% by other processes such as metal cutting).12 The main source of emissions of solvents during varnish processing are from equipment without licence requirements (about 60%) and from trade and paint work (about 15%); equipment which has licence requirements (manufacturing plants) contributes only 10% of VOC emissions.12 Therefore, solvents released during the application processes of paints and varnishes must be reduced. New EC regulations concerning reduction of solvents in special industrial plants or working processes (e.g., varnish coating of vehicles) have already been formulated.8 Similar regulations concerning other working fields (e.g., the use of solvent-based paints in home workshops) have been prepared throughout Europe and are already in force in some countries. 18.3.3.2 Immission In Europe, various regulations and schemes for the assessment of VOCs are in use according to special conditions, e.g.: • Indoor air: Schemes and values proposed by central environmental institutions commissions of UBA (Sachverständigenrat für Umweltfragen),18,19 20 (Umweltbundesamt). • Place of work: Special values (e.g., TLV-, MAK-, TRK-values) proposed by special institutions, e.g., commissions of ACIGH (American Conference of Governmental Industrial Hygienists)21 or DFG (Deutsche Forschungsgemeinschaft).22 TLV (Threshold Limit Values) refer to airborne concentrations of substances. They represent conditions to which workers may be repeatedly exposed during an 8-hour workday in a 40-hour week. MAK-values (Maximale Arbeitsplatzkonzentration) describe maximum concentrations of individual substances allowed in the work place (8 hours per day or 40 hours per week with some exceptions allowed). MAK-values are obligatory limits in Germany but in cases where MAK values cannot be evaluated (e.g., carcinogenic compounds) TRK values (Technische Richtkonzentrationen) are used. These recommendations are based on current technical knowledge.22 • Atmosphere, especially the significance of photooxidation. Various measures and regulations (e.g., reduction of VOCs)8 intend to limit substances which play a part in photooxidizing processes (e.g., ozone formation). When defining threshold limit values, it is important that prevailing conditions and methods are specified, e.g., ventilation rates, methods of sampling, determination and calculation.1,17,23


1243

Immission processes of solvents from paints and varnishes are determined by surrounding conditions: e.g., parameters of the ambient air, indoor air parameters, e.g., ventilation (ventilation rates), air distribution, air movements, composition of ambient air, concentrations of air compounds, humidity, temperature, volume of the indoor air, extreme conditions (e.g., worst case). 18.3.4 EFFECTS ON HEALTH OF SOLVENTS FROM PAINTS AND VARNISHES Only the effects resulting directly from the exposure to paints and varnish solvent are discussed in this chapter. For the effects of other solvents, see Chapter 20.1. 18.3.4.1 Exposure Usually the effects of solvents in paints and varnishes on health are dependent on concentration and exposure time. Adverse health effects may follow exposure to paints, varnishes and their solvents at the workplace. The conditions at paint and lacquer manufacturing sites (e.g., manufacturing methods, use of exhaust hoods, etc.) are responsible for the levels of evaporated solvents measured in the air.24 Adverse health effects depend on how the paint is applied, paint properties and working conditions (e.g., increased risk with spray painting). There is sufficient evidence to substantiate the fact that solvents to which painters have been exposed, are responsible for incidences of cancer. However, within paint manufacturing plants, this evidence is inadequate.25 Nevertheless, person working at home, occupants of painted rooms and children of parents which have been exposed are also at risk. Those involved with the abuse of solvents as a psychoactive substance (e.g., aromatic hydrocarbons in spray paints, mainly semi-volatile or nonvolatile components) are at a much higher risk. The main path of entry of solvents from paints and varnishes to the body is by inhalation. Volatile paint compounds present a particularly high risk as do some forms of paint application (e.g., spray painting with the risk of inhalation of even less volatile and nonvolatile paint components). Other pathways should, however, also be considered as dermal contact. In addition to solvents, other compounds from paints and varnishes can cause various diseases, often similar to the effects caused by solvents (e.g., asbestos as paint filler or in the construction and shipyard industry, silica, dusts, thermal decomposition products, contaminations of solvents, chromium, iron and lead compounds in paint pigments). It is often difficult to associate a particular components of paints and varnishes with adverse health effects. In most cases, the paint and varnish products were found to be a relevant cause of illness, but their individual compounds were not. 18.3.4.2 Health effects 18.3.4.2.1 Toxic responses of skin and mucose membranes The following symptoms involving the skin and mucose membranes may occur as a result of using paints and varnishes: • Irritations of skin and mucose membranes • Allergic diseases of skin and mucose membranes • Removal of grease from skin (removal of sebaceous matter, with subsequent adverse skin conditions such as infection) • Changes in the lens of the eye and corneal changes • Absorption of solvents through the skin (e.g, benzene, toluene, xylene, methyl alcohol, methyl ethyl ketone, glycol ethers)

1244


Workers in the paint manufacturing industry and painters have experienced occupational diseases, especially dermatosis affecting the hands and arms.26-28 Whereas several paint components have been shown to cause non-allergic and allergic contact eczema, organic solvents were shown to provoke mainly non-allergic contact eczema and some solvents cause only irritation (e.g., some ketones and esters). In another study by Mancuso et al.28 it was shown that the occupational contact dermatitis of workers in shoe factories was probably a result of solvents in varnishes and adhesives. This study was based on interviews, medical examinations and patch test series. Eye diseases stemmed from both non-allergic and allergic reactions, and in some cases corneal and lens changes were noted.29 In a further study,30 with water based paints there was a significant reduction in eye and skin disease and worker discomfort on the job. 18.3.4.2.2 Neurological disorders Indoor air immissions of organic solvents from paints and varnishes can cause neurological disorders: • Neuro-psychological and neuro-behavioral symptoms (e.g., subjective symptoms, multiple chemical sensitivity - MCS) • Neurophysiological symptoms • Neurological diseases (e.g., polyneuropathy) • Neuropsychiatric diseases Other compounds in paints and varnishes apart from solvents can cause neurological disorders (e.g., lead). However, it is difficult to prove that solvents specifically cause neurological changes. A study of production plants producing dyes and varnishes,31 showed that mixtures of organic solvents are responsible for several neurological and neurophysiological symptoms: headache, dizziness, increased emotional excitability, memory and concentration disorders, mood instability, fatigue. Neurological examinations, however, showed no significant changes in the central and peripheral nervous system but EEG and VEP anomalies were seen. In contrast to the studies mentioned above, a study of shipyard spray painters32 exposed to xylene and mixed organic solvents described neurophysiological changes, e.g., decreased nerve function and, in addition, neuropsychological symptoms, e.g., mood changes and fatigue. Similar results including reduced nerve conduction were shown by workers exposed to styrene. Other studies found several dose-response relationships between solvent mixtures and neuro-behavioral effects among paint manufacturing employees and painters.33,34 Significant relationships concerning the total amounts of hydrocarbons, lifetime exposure and lifetime-weighted average were described. In earlier reviews and cross-sectional studies, various symptoms and neuro-behavioral effects were described for workers in the paint manufacturing industry, house painters, car and industry painters, and shipyard painters.35 Subjective symptoms (fatigue, loss of concentration, emotional instability, short-term memory disorders, headache) or effects on psychomotoric performance are examples of these symptoms. However, similar former studies did not find symptoms in house painters using mainly water-based paints.36 Different results were found concerning neuro-physiological changes and neurological diseases. Electroencephalographic changes and a slight decrease in cerebral blood flow of paint industry workers was noted by Oerbaek et al.37 and there were occasional cases of


1245

clinical polyneuropathy in spray painters exposed mainly to methyl n-butyl ketone as described by Mallov.38 Furthermore, slight neurological impairments of car and industry painters was noted by Elofsson et al.29 and Maizlish et al.39 and in house painters by Askergren et al.36 Nevertheless, no similar effects were described in other studies.42,43 In particular, no effects on the peripheral nervous system were reported in workers using water-based paints.36 Inconsistent results were reported for severe diseases of the central nervous system (neuropsychiatric diseases, encephalopathy) for painters and other persons exposed to solvents.42-44,47 Environmental exposure to organic solvents is supposed to be one cause of multiple chemical sensitivity which characterizes neuropsychological disorders. Organic solvents from paints seem to play a role.45,46 The relevant exposure characteristics (dose, time, possible dose-response relationship) are also discussed.45 18.3.4.2.3 Carcinogenic effects Various solvents which occur in paints and varnishes exhibit a carcinogenic potential. Benzene and all isomers of dinitrobenzenes are particular examples (see also Chapter 20.1). This was confirmed by several extensive studies of occupational exposures in paint manufacturing plants and in painters.25 These data show the risk of contracting cancer to be about 20% above the national average. In many studies, increased risks were described mainly for lung cancer (about 40% above the national average), leukemia, bladder cancer, liver cancer, and childhood cancers where there has been parental exposure.25 In a meta-analysis using standardized mortality ratios the relationship between painting exposure48 and cancer mortality showed the highest risks for leukemia and liver cancer. Elevated risks were also predicted for lung cancer, oesophageal cancer, stomach cancer and bladder cancer. The development of leukemia, especially in the case of benzene mixed with other organic solvents, being most prominent and the development of lung cancer (main risks being lead chromate and asbestos) to a lesser degree. It should be noted, however, that more critical risk factors such as smoking and alcohol can obscure such correlations. Although high risks for cancer resulting from occupational exposures were mentioned, no significant information concerning the occupation of painted rooms could be noted. This area has been inadequately studied. As mentioned above, lung cancer is a major concern. Painters, as opposed to those involved in paint manufacture were shown to be at greatest risk for contracting lung cancer.25,48 In fact, there was no evidence of increased risk in persons involved in the manufacture of paints.25,49 Other cancers of the respiratory tract are documented also, e.g., cancer of the nasal cavity,50 pleural mesothelioma with high incidences in painters and paper-hangers51 and cancers of the larynx.52 The evidence of several types of leukemia is convincing with regard to occupational exposures to paints and solvents.25,48 Increased risks for contracting other haematopoietic neoplasmas25 such as Hodgkin´s disease,53 non-Hodgkin´s lymphoma,54 multiple myeloma,55 reticulum-cell sarcoma and lymphosarcoma56 were also reported. Probably all neoplasmas were caused mainly by organic solvents. Some former studies, however, included solvents which are rarely used now, especially benzene.25

1246


Davico et al.57 and Crane58 showed clonal aberrations in chromosome 8 of workers exposed to paints which is correlated with higher risks of acute myelocytic leukemia. Occupational exposure to dyestuff manufacture, paints, solvents and inks are an important risk factor for cancer development of the urinary tract (especially aromatic amines in bladder cancer).25,59 A major risk factor responsible for about 50% of bladder cancer cases in western countries is smoking. A second risk factor is occupational exposure (including exposure to paints and solvents).59 Some studies showed greater risks according to certain exposures: e.g., spray painting,60,61 lacquering and painting of furniture and cars or sign-post painting.62 An increased risk of cancer of the prostate was found for workers in paint manufacturing plants, but no increase in risks could be evaluated for exposure to paints.63 In one study an increased risk for testicular cancer was shown.25 Inconsistent results were noted for cancers of the gastrointestinal tract.25 Some studies showed higher rates of stomach and intestinal cancers, but other risk factors apart from paints cannot be excluded. Higher risks for liver cancer was found by Chen and Seaton.48 Norell et al.64 noted an excess of pancreatic cancers, especially as a result of exposure to paint thinners. No significant risks were described concerning cancer of the biliary tract.65 A distinct relation between parental occupational exposure and childhood cancer was shown for solvents and paints. High parental exposure resulted in higher incidences of childhood cancers.66 In the same study, however, generally more cancers were found as a result of parental use of alcohol and tobacco smoke. Childhood leukemia and nervous system cancers, in particular, are the types suspected to be caused by parental exposure to paints and solvents.67 Kishi et al.68 described an elevated risk for acute lymphatic leukemia in children of mothers with prenatal exposures to benzene and to paints. In former studies with small numbers of children these tendencies could also be shown, mainly in male painters whose children showed a higher incidence of childhood leukemia and brain tumors.25 18.3.4.2.4 Respiratory effects The following respiratory symptoms are provoked by solvents:

• • • •

Irritations Allergic reactions Changes in lung function parameters (mainly obstructive ventilation) Pathohistological changes Irritations of the air passages were described in people exposed to fumes in paint factories.27,69 VOCs in paints can provoke respiratory symptoms (wheezing, breathlessness) in asthmatics. Conventional water-based paints with only small amounts of VOCs have also been shown to cause such symptoms, but there were no effects using VOC-free paints.70 No differences were found in the same study looking at lung function and airway responsiveness. Toxic pyrolysis products in paints and polymer films probably evoke asthma-like symptoms similar to PVC pyrolysis.71 A decrease in forced vital capacity, expiratory volume in one second and of peak exspiratory flow was observed after exposure to water-based paints.72 Bronchial obstruction in painters was confirmed by White and Baker,69 but other studies could not detect changes in lung function parameters in house painters who were exposed to solvent-based and water-based paints.36 Beving et al.73 did not find obstructive effects in car painters. An increase


1247

in mortality from chronic obstructive bronchitis of painters was recorded by Engholm and Englund,74 but not confirmed in other studies.25 Histological changes in the nasal mucosa of spray painters were also noted in a study of Hellquist et al.75 18.3.4.2.5 Toxic responses of the blood A case-control study of Guiguet et al.76 observed an increased risk for aplastic anaemia after exposure to paints (effective compounds in the paints are unknown), but no increased risk after exposure to solvents alone. Reduction of blood hemoglobin levels was shown after exposure to gasoline, car spray paint (xylene) and various solvents (house painters).25,77,78 An increased Hb level was noted, however, in a study of car and industrial spray painters.29 Alterations of several blood components (production of auto-antibodies) and of vascular endothelial cells were described for workers exposed to hydrocarbon-based paints or to mixed solvents.79 Even a group experiencing low exposure (house painters exposed to alkyd paints) showed hematological changes. Higher platelet counts and higher resistivity calculated from the the impedance of the whole blood were observed.73 Some studies showed decreases of thrombocyte counts in car painters and paint industry workers, other studies, however, recorded no changes.25 Studies concerning effects of paints on the white cell and thrombocyte counts were inconsistent. A decrease of white cells was described in several studies11 and a lymphocytosis was noted by Angerer and Wulf.77 Elofsson et al.,29 however, recorded no changes in white cell counts. Additionally, myelotoxic effects of solvents were shown, especially for benzene. 18.3.4.2.6 Toxic responses of the reproductive system In interviews with pregnant women, the effects of paints, varnishes and solvents were evaluated. No effects were found concerning congenital heart disease of infants with Down Syndrome (trisomie 21).80 The maternal risk was associated with smoking. Maternal toluene sniffing used as an organic solvent in acrylic paints, varnishes and other sources is associated with premature births and, in one case, renal tubular acidosis.81 Other studies of female painters showed tendencies towards an increase of spontaneous abortions82 or infant mortality.83 Other studies84 did not confirm these tendencies. McDowall reported an increase in malformations in children (polydactyly, syndactyly, spina bifida and anencephalus) whose parents were painters, assemblers or had related occupations.83 A study by Olson85 showed congenital malformations of the central nervous system for paternal exposure to paints. 18.3.4.2.7 Toxic responses on other organ systems Contradictory effects of solvents in paints on the kidneys have been described.25 Lauwerys et al., for example, have described some solvents used in paints as nephrotoxic (e.g., toluene).88 Minor effects such as slight hematuria and albuminuria or small effects on the glomeruli or even no effects were detected in other studies.25 In some studies of spray painters in the automobile and airplane industries and of house painters mortality from liver cirrhosis was increased.25,87 Another study investigating house painters, however, did not show an increase.42 There seem to be no increased risk for cardiovascular diseases according to some studies of paint industry workers.25,89 Nevertheless, effects of solvents on muscles and vessels could be shown (e.g., increased serum creatinine levels).25

1248


18.3.5 METHODS FOR THE EXAMINATION OF SOLVENTS IN PAINTS AND VARNISHES 18.3.5.1 Environmental monitoring There are several test procedures generally used in Europe. Procedures are still being developed to determine the presence of solvents in paints and varnishes. 18.3.5.1.1 Solvents in products Officially approved test procedures are: • Gravimetry: Determination of volatile organic compounds (DIN ISO 11 890-17 VOCs > 15 %). • Gas chromatography: Determination of volatile organic compounds (DIN ISO 11 890-27 - VOCs 0,1 % -15 %). 18.3.5.1.2 Emission of solvents Test chamber7,92,93 (volume 1 m3, defined conditions: temperature, humidity, air changing rates, air velocity). This is usually carried out 3 days after application of paints or varnishes by placing the products into a test chamber. The test series starts with sampling emissions of the products on defined absorption materials (Tenax, charcoal filters, activated charcoal), reconditioning and finally identification and classification of VOCs with gas chromatography. The test methods are repeated after 14 and 28 days. These test criteria permit determination of the behavior of emissions from finished products under defined conditions. Emission test cells.7,92,93 The product under investigation is hermetically sealed into the emission test cells. The emitted compounds are sampled by absorption materials and analyzed. This method can be used under laboratory conditions or in situ. 18.3.5.2 Biological monitoring of solvents in human body fluids 18.3.5.2.1 Solvents and metabolites in human body fluids and tissues The biological monitoring of solvents emitted from paints or varnishes on humans is not well developed. In two studies,94,95 solvents from paints and varnishes were determined in blood, urine and internal breath. Blood and urine analysis is less sensitive than internal breath measurements. This was carried out in a study on exposure to paints in aircraft maintenance.94 Kramer et al.95 found xylenes during paint production and paint-spraying in ambient air and in the blood and urine of workers. Threshold limit values (TLV) and biological exposure indices (BEI) were not exceeded.22,23 18.3.5.2.2 Biomarkers Even low air concentrations of solvents emitted from paints have an influence on the human organism, e.g. the induction of hepatic enzymes. This effect was shown in workers exposed to butyl glycol from paints in an electrophoresis painting plant where the exposure value was shown to be less than 0.3 times of the average limit. In these cases D-glucaric acid in urine, which reflects the D-glucarid acid enzyme pathway, was increased.96 In some studies the genetic effects on paint industry workers could not be detected when chromosomal aberrations and sister chromatid exchanges were studied.86,87 In other studies94,95 it was shown that the frequency of micronuclei and sister chromatid exchanges increased.


1249

REFERENCES 1 2 3

4 5 6 7 8 9

10

11 12

13 14 15 16 17 18 19 20 21 22 23

24 25

26 27 28 29

30

CEN (European Committee for Standardization), Paints and varnishes. Terms and definitions for coating materials. Part 1: General terms. EN 971, 1996. Ullmann`s Encyclopedia of Industrial Chemistry, 1998. DIN 55945 (Deutsches Institut für Normung), Coating materials (Varnishes, Painting Materials and other Materials), Terms. Beschichtungsstoffe (Lacke, Anstrichstoffe und ähnliche Stoffe); Begriffe, Beuth Verlag, Berlin, 1988. VdL (Verband der Lackindustrie/ Association of the Varnish Industry in Europe)- VdL-Guidelines Building Coating Materials. VdL-Richtlinie Bautenanstrichstoffe, RL 01, 1997. Römpp Lexikon, Varnishes and Paints. Lacke und Druckfarben, Thieme, 1998. H. Zeh, M. Kroll, and K. Kohlhammer, VDI Berichte, 1122, pp. 455-475 (1994). DIN (Deutsches Institut für Normung), ISO/ DIN 11890. Paints and varnishes - Determination of volatile organic compound (VOC) content, Beuth Verlag, Berlin, 1998. European Commission, ECA-IAQ, Total Volatile Organic Compounds (TVOC) in Indoor Air Quality Investigations, EUR 17675 EN, Report No. 19, 1997. 1999/13/EC, Council Directive of 11 March 1999 on the limitations of emission of volatile organic compounds due to the use of organic solvents in certain activities and installations. ABl. EG Nr. L 085, 29.3.99, 1999. DIN (Deutsches Institut für Normung)(Normenausschuss Bauwesen NABau), Koordinierungsausschuss KOA 03 “Hygiene, Gesundheit und Umweltschutz”: Guidelines for Health Assessment of Building Products. Leitfaden zur Beurteilung von Bauprodukten unter Gesundheitsaspekten, KRdL-CEN/264/7 N 1999-07, 1999. GISBAU, WINGIS, Bau-Berufsgenossenschaften (Professional Associations of the Building Industry in Germany), 1999. Deutsches Lackinstitut Frankfurt a.M. (German Institute of Varnishes Frankfurt a.M.), Solvent reduction in the trade of painters and lacquerers. Documents concenring varnishes and paints. Magazine No. 7. Lösemittel-Reduzierung im Maler- und Lackiererhandwerk. Dokumente zu lacken und Farben. Heft Nr. 7, DeutschesLackinstitut, Frankfurt a.M., 1999. T. Salthammer, Chemie in unserer Zeit, 28, 280 (1994). DIN (Deutsches Institut für Normung), DIN ISO 11890-1,-2, Beuth Verlag, Berlin, 1998. VDI (Verein Deutscher Ingenieure), VDI 4300, VDI, Düsseldorf, 1999. L. Molhave, Indoor Air, 1, 357 (1995). Bundesgesundheitsamt (Public Health Department of the German Government), Bundesgesundheitsblatt, 36 (3), 117-118 (1993). SRU (Environmental Council of the German Government)(Sachverständigenrat für Umweltfragen), Air pollutions in rooms, Special Report, Metzler-Poeschel, Stuttgart, 1987. SRU (Environmental Council of the German Government)(Sachverständigenrat für Umweltfragen), Environmental Report 1996, Metzler-Poeschel, Stuttgart, 1996. Anonymous, Bundesgesundheitsblatt, 11, 422 (1996). ACGIH (American Conference of Governmental Industrial Hygienists), TLVs and BEIs. Threshold Limit Values for Chemical Substances and Physical Agents. ACGIH, 1998. DFG (Deutsche Forschungsgemeinschaft), MAK- und BAT-Werte-Liste 1999, Wiley-VCH, 1999. M. Fischer, and E. Böhm, Evaluation and assessment of emission of noxes caused by furniture varnishes. Erkennung und Bewertung von Schadstoffemissionen aus Möbellacken. Schadstoffe und Umwelt, Erich Schmidt Verlag, Berlin, 1994. W. Wesolowski, and J.P. Gromiec, Int. J. Occup. Med. Environ. Health, 10 (1), 79 (1997). IARC (International Agency for Research on Cancer), IARC Monographs on the evaluation of carcinogenic risks to humans. Some organic solvents, resin monomers, resin monomers and related compounds, pigments and occupational exposures in paint manufacture and painting. WHO, 47, IARC, Lyon, 1989. U. Ulfvarson, Int. Symp. on the Control of Air Pollution in the Working Environment, Stockholm, September 6-8, Swedish Work Environment Fund/ International Labour Office, Part II, 1977, 63-75. R.V. Winchester and V.M. Madjar, Ann. Occup. Med., 4, 221 (1986). G. Mancuso, R.M. Reggiani, and R.M. Berdandini, Contact Dermatitis, 34 (1), 17 (1996). S.A. Elofsson, F. Gamberale, T. Hindmarsh, A. Iregren, A. Isaksson, I. Johnsson, B. Knave, E. Lydahl, P. Mindus, H.E. Persson, B. Philipson, M. Steby, G. Struwe, E. Söderman, A. Wennberg, and L. Widen, Scand. J. Work Environ. Health, 6, 239 (1980). G. Wieslander, D. Norback, and C. Edling, Occup. Environ. Med., 51(3), 181 (1994).

1250

31 32 33 34 35 36 37 38 39 40 41

42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71


J.A. Indulski, H. Sinczuk-Walczak, M. Szymczak, and W. Wesolowski, Int. J. Occup. Med. Environ. Health, 9(3), 235 (1996). M.W. Ruijten, J.T. Hooisma, J.T. Brons, C.E. Habets, H.H. Emmen, and H. Muijser, Neurotoxicology, 15 (3), 613 (1994). F. Fung, and R.F. Clark, J. Toxicol. Clin. Toxicol., 37(1), 91 (1999). A. Seeber, B. Sietmann, and M. Zupanic, Int. J. Occup. Med. Environ. Health, 9 (3), 235 (1996). NIOSH (National Institute for Occupational Safety and Health), Organic solvent neurotoxicity, Current Intelligence Bulletin No. 48, Cincinnati, 1987. A. Askergren, H. Beving, M. Hagman, J. Kristensson, K. Linroth, O. Vesterberg, and A. Wennberg, Arb. Hälsa, 4, 1 (1988). P. Oerbak, J. Risberg, I. Rosen, B. Haeger-Aronsen, S. Hagstadius, U. Hjortsberg, G. Regnell, S. Rehnström, K. Svensson, and H. Welinder, Scand. J. Work Environ. Health, 11 (Suppl. 2), 1 (1985). J.S. Mallov, MBK neuropathy among spray painters, J. Am. Med. Ass., 235, 1455 (1976). N.A. Maizlish, G.D. Langolf, L.W. Whitehead, L.J. Fine, J.W. Albers, J. Goldberg, and P. Smith, Br. J. Ind. Med., 42, 579 (1985). N. Cherry, H. Hutchins, T. Pace, H.A. Waldron, Br. J. Ind. Med., 42, 291 (1985). D. Triebig, D. Claus, I. Csuzda, K. Druschky, P. Holler, W. Kinzel, S. Lehrl, P. Reichwein, W. Weidenhammer, W.U. Weitbrechst, D. Weltle, K.H. Schaller, and H. Valentin, Int. Arch. Occup. Environ. Health, 60, 223 (1988). S. Mikkelsen, Scand. J. Soc. Med. Suppl., 16, 34 (1980). H. Rasmussen, J. Olsen, and L. Lawitsen, J. Occup. Med., 27, 561 (1985). C. Van Vliet, G. Swaen, J. Slangern, T. Border, F. Stirman, Int Arch. Occup. Environ. Health, 59, 493 (1987). E. Kiesswetter, Zbl. Hyg. Umweltmed., 202, 191 (1998). N. Fiedler and H. Kiepen, Environ. Health Perspect., 105 Suppl. 2, 409 (1997). S. Mikkelsen, M. Jorgensen, E. Browne, and C. Gyldensted, Acta Neurol. Scand. Suppl., 118, 1(1988). R.L. Chen, and A. Seaton, Cancer detection and prevention, 22 (6), 533 (1998). K.L. Milne, D.P. Sandler, R.B. Everson, S.M. Brown, Am. J. Ind. Med., 4, 565 (1983). S. Hernberg, P. Westerholm, K. Schultz-Larsen, R. Degerth, E. Kuosma, A. Englund, U. Engzell, H. Sand Hansen, and P. Mutanen, Scand. J. Work Environ. Health, 9, 315 (1983). H.S.R. Malker, J.K. McLaughlin, B.K. Malker, B.J. Stone, J.A. Weiner, J.L.E. Erickson, and W.J. Blot, J. Natl. Cancer Inst., 74, 61 (1985). L.M. Brown, T.J. Mason, L.W. Pickle, P.A. Stewart, BP. Auffler, K. Burau, R.G. Ziegler, and F. J. Fraumeni, Cancer Res., 48, 1960 (1988). H. Olssen, and L. Brandt, Scand. J. Work Environ. Health, 14, 246 (1980). H. Olssen, L. Brandt, Scand. J. Work Environ. Health, 6, 302 (1988). J. Cuzik, and B. De Stavola, Br. J. Cancer, 57, 516 (1988). N.J. Vianna, and A. Polan, Lancet, 1394 (1979). L. Davico, C. Sacerdote, G. Ciccone, L. Pegoraro, S. Kerim, G. Ponzio, and P. Vineis, Cancer Epidemiol. Biomarkers Prev., 7 (12), 1123 (1998). M.M. Crane, Cancer Epidemiol. Biomark. Prev., 5, 639 (1996). R. Piratsu, I. Iavarone, and P. Comba, Ann. Ist. Super Sanita, 32 (1), 3 (1996). J. Claude, E. Kunze, R. Frentzel-Beyme, K. Paczkowski, J. Schneider, H. Schubert, Am. J. Epidemiol., 124, 578 (1986). J. Claude, R. Frentzel-Beyme, E. Kunze, Int. J. Cancer, 41, 371-379 (1988). O.M. Jensen, J. Wahrendorf, J.B. Knudsen, B.L. Sorensen, Scand. J. Work Environ. Health, 13, 129 (1987). R.C. Brownson, J.C. Chang, J.R. Davis, and J.R. Bagby, J. Occup. Med., 30, 523 (1988). S. Norell, A. Ahlbom, R. Olin, R. Ewald, G. Jacobsen, I. Lindberg,-Navier, and K. Wiechel, Br. J. Ind. Med., 43, 775 (1986). H.S.R. Malker, J.K. McLaughlin, B.K. Malker, B.J. Stone, J.A. Weiner, J.L.E. Erickson, and W.J. Blot, Br. J. Ind. Med., 43, 257 (1986). M.L. McBride, Can. J. Public Health, 89 (1), 53 (1998). J.S. Colt, and A. Blair, Environ. Health Perspect., 106 (3), 909 (1998). R. Kishi, Y. Katakura, J. Yuasa, and H. Miyake, Sangyo Igaku, 35 (6), 5115 (1993). M.C. White and E.L. Baker, Br. J. Ind. Med., 45, 523 (1988). J.R. Beach, J. Raven, C. Ingram, M. Bailey, D. Johns, E.H. Walters, and M. Abramson, Eur. Respir. J., 10 (3), 563 (1997). J.E. Peterson, Occup. Med., 8(3), 533 (1993).

18.4 Solvent uses with exposure risks

72 73 74

75 76 77 78 79 80 81 82 83

84 85 86 87 88 89 90 91 92 93 94 95 96

1251

U. Ulfvarson, R. Alexandersson, M. Dahlqvist, U. Ekholm, B. Bergstrom, and J. Scullman, Scan. J. Work. Environ. Health, 18 (6), 376 (1992). H. Beving, B. Tedner, and L.E. Eriksson, Int. Arch. Occup. Environ. Health, 63 (6), 383 (1992). G. Engholm and A. Englund in Advances in Modern Environmental Toxicology, Vol. II, Occupational Health Hazards of Solvents, A. Englund, K. Ringen, and M.A. Mehlman, eds., Princeton Scientific Publishers, Princeton, N.J., pp. 173-185 (1982). H. Hellquist, K. Irander, C. Edling, and L.M. Ödkvist, Acta Otolaryngol., 96, 495 (1983). M. Guiguet, E. Baumelou, and J.Y. Mary, Int. J. Epidemiol., 24 (5), 993 (1995). J. Angerer, and H. Wulf, Int. Arch. Occup. Environ. Health, 56, 307 (1985). M. Hane, O. Axelson, J. Blume, C. Hogstedt, L. Sundell, B. Ydreborg, Scand. J. Work Environ. Health, 3, 91 (1977). A. Stevenson, M. Yaqoob, H. Mason, P. Pai, and G.M. Bell, QJM, 88 (1), 23 (1995). D.E. Fixler, N. Threlkeld, Teratology, 58 (1), 6 (1998). J. Erramouspe, R. Galvez, and D.R. Fischler, J. Psychoactive Drugs, 28 (2), 201 (1996). L.Z. Heidam, J. Epidemiol. Commun. Health, 38, 149 (1984). M.E. McDowall, Occupational Reproductive Epidemiology: The Use of Routinely Collected Statistics in England and Wales, 1980-82, Studies on Medical and Populations Subjects No. 50, Her Majesty´s Stationary Office, Office of Population Censuses and Surveys, London, 1980. W.E. Daniell, and T.L. Vaughan, Br. J. Ind. Med., 45, 193 (1988). J.H. Olsen, Dan. Med. Bull., 30, 24 (1983). U. Haglund, I. Lundberg, and L. Zech, Scand. J. Work Environ. Health, 6, 291 (1980). K.T. Kelsey, J.K. Wiencke, F.F. Little, E.L. Baker, J.B. Little, Environ. Mol. Mutagenesis, 11, 389 (1988). R. Lauwerys, A. Bernard, C. Viau, and J.P. Buchet, Scand. J. Work Environ. Health, 11 (1), 83 (1985). G. Engholm, A. Englund, H. Löwing, Scand. J. Work Environ. Health,. 13,. 181 (1987). R.W. Morgan, K.W. Claxton, S.D. Kaplan, J.M. Parsons, and O. Wong, J. Occup. Med., 27, 377 (1985). I. Lundberg, Scand. J. Work Environ. Health., 12, 108 (1986). European Collaborative Actions (ECA), Indoor Air Quality & Ist Impact On Man. Evaluation of VOC Emissions from Building Products - Solid Flooring Materials, ECA Report No. 18, 1997. Anonymous, Building products - Determination of volatile organic compounds Part 1: Emission test chamber method. prENV 13419-1, draft. G.K. Lemasters, J.E.Lockey, D.M. Olsen, S.G. Selevan, M.W. Tabor, and G.K. Livingston, Drug and Chemical Toxicology, 22 (1), 181 (1999). A. Kramer, M. Linnert, R. Wrbitzky, and J. Angerer, Int. Arch. Occup. Environ. Health, 72(1), 52 (1999). J.P. Collinot, J.C. Collinot, F. Deschamps, D. Decolin, G. Siest, and M.M. Galteau, J. Toxicol. Environ. Health, 48 (4), 349 (1996).

18.4 SOLVENT USES WITH EXPOSURE RISKS Pentti Kalliokoski University of Kuopio, Kuopio, Finland

Kai Savolainen Finnish Institute of Occupational Health, Helsinki, Finland

18.4.1 INTRODUCTION This chapter deals merely with exposure to common organic solvents which are used in large quantities to dissolve fats, resins, and other materials. Very dangerous chemicals, such as benzene, which are no more used as solvents due to their toxic properties, will not be discussed even though they may have had even extensive uses as solvents earlier and even though those may be still important chemicals as petroleum components or as intermediates for other chemicals. Also, solvents with very specialized uses, such as carbon disulfide the use of which is practically limited to viscose rayon industry and laboratories, are only

1252

Pentti Kalliokoski, Kai Savolainen

shortly considered. Exposure data presented originate mainly from literature; however, some unpublished data obtained from the Finnish Institute of Occupational Health (FIOH) has been added to indicate the order of magnitude of current exposure levels in industrialized countries (data cover the years 1994-1996). Exposure to organic solvents should be avoided mainly due to the risk of neurotoxic effects. Acute effects are narcotic resembling those caused by the use of alcohol. Those appear, for example, as a decrease in reaction time and impairment in psychological performance. Chronic neurotoxic effects are often called as the organic psychosyndrome (OPS) including memory disturbances, excessive tiredness, personality changes, irritability and affect lability. Intellectual reduction may occur but it seems that development of real dementia would require simultaneously heavy consumption of alcohol.1 Increased risk of sleep apnoea has been observed among men exposed occupationally to organic solvents.2 Exposure to toluene, xylene, and styrene has been found to contribute to the development of noise-induced hearing loss. This interaction has been suggested to be due to neurotoxic injuries caused by solvents in the cochlea.3 Long-term exposure to carbon disulfide and n-hexane may result in peripheral neuropathy. It should be noted that n-heptane and n-octane have not been shown to cause the type of peripheral neuropathy (numbness, weakness, and pain in extremities) associated with n-hexane.4 Otherwise than the other alkanes, n-hexane is metabolized to a reactive, toxic compound (2,5-hexanedione).5 Organic solvents are generally skin irritants. Repeated or prolonged contact may cause erythema and dryness of the skin. Defatting may lead to cracked skin. Many solvents also readily absorb through the skin (see 18.4.2). Reactive solvents, such as styrene and vinyltoluene, may cause contact dermatitis.6 The annual incidence of occupational diseases caused by organic solvents was 20.6 cases/10,000 exposed workers in Finland in the 1980’s. More than half of the cases (64%) were dermatoses.7 Benzene is a well-known human carcinogen. An association with exposure to benzene and leukemia was detected already in 1920’s.8 There is some evidence that exposure to other organic solvents may constitute a carcinogenic risk. The International Agency for Research on Cancer (IARC) has concluded that there is sufficient evidence for the carcinogenicity of painters’ occupational exposure.9 Epidemiological studies conducted among dry cleaning and metal degreasing workers suggest an increased cancer risk due to exposure to chlorinated solvents.10 The carcinogenicity of aromatic solvents (styrene, toluene, and xylene) has been investigated in many large epidemiological studies. Although the results have been inconclusive, certain site-specific associations (e.g., for lymphohematopoietic tissues) have appeared in some of the studies.11 Only styrene has shown to have some genotoxic activity in animal studies.9 18.4.2 EXPOSURE ASSESSMENT Inhalation is usually clearly the most important route for occupational exposure to solvents. Organic solvents also enter the body through the skin.12 For some solvents, especially for alkoxyalcohols (glycol ethers) and their acetates, this is even the main route of absorption. Even the skin uptake of vapor can be significant for these compounds.13 Dermal absorption of vapors is, however, usually negligible and contact with liquid is required. Increased workload and heat enhance both inhalation and dermal absorption.14 Skin contamination may also lead to oral uptake due to eating and smoking but this is of minor importance for volatile solvents.


1253

Most industrial countries have occupational exposure limits (OELs) for airborne concentrations to prevent excessive exposures. The threshold limit values (TLVs) published by the American Conference of Governmental Industrial Hygienists (ACGIH15) are unofficial but have had a great impact on OELs in the Western countries. These are updated annually based on new epidemiological and toxicological data. Organic solvents are often used as mixtures. Because the OELs for most solvents base on their neurotoxic properties, their effect is considered to be additive and the combined exposure levels (or hygienic effects) are calculated as the sum of the fraction of the OEL that each solvent represents. If the sum is larger than one, a noncompliance situation exists. Compounds which may be dangerous by uptake through the skin carry a skin notation. However, no quantitative dermal exposure limits are available yet. Methanol, turpentine, glycol ethers, and many chlorinated and aromatic hydrocarbons have the skin notation. Exposures can also be assessed by analyzing biological specimens, such as blood or urine. Metabolites of compounds are usually analyzed from urine samples. The amount of biological exposure limits or indices (BEIs) is, however, much smaller than that for airborne concentrations. However, biological monitoring may be useful although no BEI has been established to ascertain effectiveness of personal protection or to follow exposure trend. Biological monitoring reflects exposure via all routes. This is beneficial for individual exposure assessment. Biological monitoring also reveals possible accumulation of a compound. This is, however, of minor importance for most solvents because of their short half-lives. The largest disadvantage of biological monitoring is that it does not provide any information on the reasons of exposure. About twenty solvents have an ACGIH BEI (e.g., acetone, carbon disulfide, 2-ethoxyethanol, n-hexane, methanol, methyl ethyl ketone, perchloroethylene, styrene, toluene, trichloroethylene, and xylene).15 Airborne concentrations of solvents are most commonly determined by taking samples on adsorbents, such as activated carbon. Sampling can be done with a pump or passively based on diffusion. In industrial working places where airborne concentrations of solvents are on ppm-level, samples are generally extracted with a solvent (carbon disulfide is the most common one). In offices and other nonindustrial environments where concentrations are at ppb-level, samples are taken onto Tenax adsorption tubes which are desorbed thermally. In both cases, gas chromatography is the most common analytical method. Because the sampling time generally varies, it is important to calculate the time-weighted average (TWA) concentration. TWA concentration is obtained by using the sampling times as weights. Sampling and analysis has been reviewed e.g. by Soule.16 Direct-reading instruments, such as infrared or photoionic analyzers are good if only one solvent is present but interferences may be a problem when solvent mixtures are analyzed. Monitoring methods for dermal exposure have not been standardized as well as the methods to assess inhalation exposure. Adsorptive pads of activated carbon cloth can be used for monitoring of dermal exposure to organic solvents. Patches are attached on various parts of the body under the clothing and the amounts analyzed are then multiplied with the areas of each body region.17 Exposure assessment is complicated by the contribution of solvent vapor on the samples. Because the adsorptive surface of activated carbon is much larger than that of the skin, pads will adsorb much more solvents than the skin. However, only exposure to liquid phase is usually significant; therefore, the adsorbed vapor should be subtracted from the total mass analyzed.18 However, this makes the exposure assessment

1254


quite complicated, and if biological monitoring methods are available those usually provide a more practical alternative. Urine and blood are the most commonly used biological samples. Exhaled air samples can also be used. Sampling is usually carried out at the end of the shift. Because the amount of water consumed affects the concentration of the solvent or its metabolite in urine, creatinine (a normal constituent of urine) correction is often applied (mass analyzed is presented per gram of creatinine). Alternatively, the correction can be made for the relative density of urine. A density of urine of 1.024 is usually applied for this purpose. In principle, collection of 24-hour urine would be an ideal approach for biological monitoring because then the actually excreted amount of the metabolite would be measured. Because this is very difficult to carry out creatinine or density correction is generally chosen as a more practical alternative. Unfortunately, not only several ways of correction for the density of urine are being used but also several different units have been adopted. For example, the ACGIH BEIs are usually given as mg/g creatinine but the values of the FIOH and many other European organizations are presented as µmol/l of urine. In addition, other units, such as µmol/mol of creatinine, are used. The mass units given in mg can be converted to mmol by dividing by the molecular weight of the compound (molecular weight of creatinine is 113). On the other hand, because the concentration of creatinine in urine varies and it is determined separately for each individual sample, values given with and without creatinine correction (or density correction) cannot be compared directly. That is the reason why the BEIs given by different institutions or agencies have not been converted to the same units in the text. As a rule of thumb, the concentration per gram of creatinine can be obtained by multiplying the concentration per liter of urine with a factor of 0.5 - 1.0. Airborne solvent concentrations usually vary much with time. Although repeated random personal sampling is theoretically the optimal method for inhalation exposure assessment, it is very time-consuming and does not necessarily reveal the reasons for exposure. Often, occupational hygiene surveys are conducted only to make certain that the concentrations of air impurities are in compliance with the OELs. Also, the European standard (EN 689/95) for occupational exposure assessment is primarily issued for this purpose. It would, however, be practical if the measurements conducted would also provide useful information for planning of remedial measures if the measurements reveal those to be necessary. For this purpose, it is important to recognize the difference between manual tasks and process industries. In manual tasks, emissions are released very close to the worker. Most problematic solvent exposures occur while performing manual tasks, such as painting, gluing, degreasing, and cleaning. Batch processes in paint and printing ink manufacture also contain many manual tasks. Manual lay-up methods are common in the reinforced plastics industry. The workers often perform different tasks with different exposure levels. All major tasks should be investigated under various conditions. Smoke tube tests provide useful information on the spreading of solvent vapor and the efficiency of local exhausts. Good enclosure for the emission source is important for successful exposure control. Detailed instructions for ventilation arrangements in various industries are available (references will be given later). Rotogravure printing is an example of a process industry with solvent exposures. The process is, in principle, closed but emissions take place from the openings for the paper web. The workers do not need to stay in the immediate vicinity of the emission sources but can


1255

often spend most of their time in the clean control room. Thus, the workers’ exposure levels depend mainly on their moving pattern during the work shift. In addition, the emission rate depends on the production rate, i.e. on the consumption rate of the solvent (toluene). Stationary sampling can be applied for exposure assessment when it is combined with questionnaire on use of time in various areas. This can be done with a direct-reading instrument because the airborne concentration level remains quite stable.20 The possible skin exposure should also be taken into consideration. If skin contamination seems possible adequate protective clothing should be used. Glove selection is an important but difficult issue. Glove materials often tolerate organic solvents poorly. Glove manufacturers have useful information for selecting gloves for individual solvents. Especially difficult is, however, to find gloves protecting efficiently penetration of solvent mixtures. Nitrile (butadiene-acrylonitrile copolymer) gloves are often chosen in such a situation. There is, however, considerable intermanufacturer and even batch lot variability in penetration of solvents through nitrile gloves.19 The workers should also be instructed not to use thinners for hand and skin cleaning. Total quality management (TQM) is an effective way to ensure also a high quality of the working environment. The guidance for right and safe working practices should be subjoined to all working instructions. Employee participation is an essential feature of a well working TQM, and also greatly assists the achievement of the hygienic goals set. 18.4.3 PRODUCTION OF PAINTS AND PRINTING INKS Painting technology has changed over the years. The exposure levels were generally highest between the mid-1950’s and mid-1960’s when solvent-based paints were used extensively both as construction and industrial paints, and exposure control technology was still undeveloped.21 Rotogravure and silkscreen printing inks contain organic solvents. Until 1950’s benzene was used as the solvent in rotogravure inks and it remained as an impurity in toluene until 1960’s.22 Today, toluene used in rotogravure inks does not contain benzene. Alkoxyalcohols and their acetates are used as silkscreen ink solvents. The development of safer solvents has been started quite recently. First, alkoxyethanols were replaced by their acetates which have lower vapor pressures. This did not, however, improve the safety much because the skin is the main route of absorption for both alkoxyethanols and alkoxyethylacetates, and they are considered to be equally toxic. Their substitution by alkoxypropanols and their acetates is, however, a significant improvement because those do not metabolize to toxic alkoxyacetic acids.23 High exposures remained common in paint and printing ink industry still in 1970’s; for example, most solvent measurements conducted in these industries in Finland exceeded the present OELs.24 Water-based paints are today clearly most common in construction painting. Alkyd-based construction paints with white spirit (Stoddard solvent, mineral spirit or solvent naphtha) as a solvent are, however, still produced. Solvent-based paints have remained most common in industrial painting, even though the solventless powder paints are also produced in large quantities. The main products are nowadays manufactured in automated processes. On the other hand, special products are also usually made in batches and include several manual tasks. In addition to paints and inks, thinners are often canned. When the processes are provided with proper enclosures and local exhausts, the airborne solvent concentrations can be kept well below the OELs. An easy but important control measure is to keep all solvent containing pots covered. Xylene (TLV 100 ppm, the Finnish OEL 100 ppm with skin notation), toluene

1256


(TLV and the Finnish OEL 50 ppm with skin notation), butanol (TLV 50 ppm as ceiling value and with skin notation, the Finnish OEL 50 ppm with skin notation), and white spirit (TLV 100 ppm, the Finnish OEL 770 mg/m3 or 135 ppm) are usually the main concerns in paint manufacturing plants.24,25 In the 1980’s, the combined solvent exposure levels were generally below the OEL but high exposure levels were detected in pot cleaning.26 The situation has further improved to some extent in Finland in 1990’s. The measurements conducted in a large Finnish paint factory in 1996-7 yielded the following mean combined personal solvent exposure levels: batch production 0.34 (range 0.14 -0.72), automated production lines 0.17, filling 0.41 (range 0.14-1.0), and pot cleaning 1.51 (range 0.28-3.06). Thus, pot cleaning remains as a problematic task. Exposures to xylene and 1-methoxy-2-propanol caused the largest contributions in the hygienic effects. Xylene was the solvent with the highest airborne concentration (mean 26 ppm; range 0.1-359 ppm) compared to the OELs in the Finnish paint manufacturing plants in general in the mid 1990’s. It has been reported that no profound changes in exposures have occurred in the Swedish paint industry since the mid-1980’s.21 In the Finnish printing ink plants, the combined solvent levels were still in the 1980’s often out of compliance when compared to the OELs.26 Significant airborne concentrations were observed for toluene, ethyl acetate (TLV 400 ppm, the Finnish OEL 300 ppm), aromatic solvent naphtha ( the Finnish OEL 240 mg/m3), and acetone (TLV and the Finnish OEL 500 ppm). The cleaning of vessels of barrels was again an especially problematic task. If cleaning is done manually, it is difficult to control the exposure sufficiently well with local ventilation but respiratory protection is also needed. 18.4.4 PAINTING Painters are probably the largest worker group exposed to solvents. This may also be the reason why much data on occupational health risks due to solvent exposure originate from painting work although the exposure levels have been generally lower than e.g. in paint manufacture. On the other hand, chronic neurotoxic effects and especially cancer require long exposure time (more than 10 years) and thus the patients with solvent-related disorders have been exposed long ago when the exposure levels, especially in construction painting, were considerably higher than nowadays. In addition, a solvent induced mild toxic encephalopathy is often a progressive disease after the cessation of exposure. There is, however, a great individual variability both in susceptibility and prognosis of the disease.27 Several studies have indicated an interaction between solvent exposure and high alcohol consumption. It has even been suggested that solvent exposure and use of alcohol are acting synergistically.28 The changes in paint technology have had a great impact on exposure to solvents in construction painting where the influence has been much more significant than in paint production because of poor ventilation. Period of high solvent exposure lasted from mid-1950’s to mid-1970’s.28 At that time, even acute intoxications occurred. Painting and lacquering using epoxyester formulations often caused very high exposure levels.24 The average exposure level to white spirit was estimated to be 130 ppm for painters when alkyd paints were extensively used. During actual painting situations, the exposure levels could rise to 300 ppm. Levels of about 200 ppm are, however, more typical in conditions of poor ventilation. If doors and windows can be kept open, the airborne white spirit concentration is significantly reduced, about to 40 ppm. In early 1980’s, the average exposure levels to white spirit were reduced to about 40 ppm in Finland.29 Nowadays, water-based paints are


1257

also commonly used in repair and maintenance painting and the average exposure levels have become generally low.28 Occasionally, high peak exposures continue to take place when solvent-based paints are used in poorly ventilated spaces. When small items are painted industrially, the worker can stay outside the spray booth and exposure to solvents can be effectively controlled. Measurements conducted in the 1970’s already indicated that no serious problems existed in that kind of work.29 Video terminal painting is a modern example of such a situation.30 On the other hand, when large metal products, such as cars and trailers, are painted, the worker must enter the booth and control is much more difficult. Solvents are not, however, the biggest hygienic problem in car painting but the paint mist (overspray). The measurements carried out in Finland in the 1980’s indicated that the combined solvent concentration was usually less than 50% of the OEL.26 On the other hand, the paint mist concentrations exceeded clearly the standards.29 If painting is done in a poorly ventilated booth, high solvent concentrations, however, appear.30 Isocyanate-based urethane paints are nowadays common in car painting. They contain 1,6-hexamethylene diisocyanate (HDI) based polyisocyanates, which have become a major cause of occupational diseases, especially asthma.31,32 Downdraft spray-painting booths provide the best overspray control. Air velocity, flow direction, and flow homogeneity are the ventilation parameters having the largest effect on booth performance. Spray-painting booths do not, however, completely control exposure to paint mist and isocyanates.33,34 In addition to actual painting, the workers may become exposed to solvents during other tasks, such as solvent cleaning. For example, isopropanol (TLV 400 ppm, the Finnish OEL 200 ppm) is used for wiping cars before the application of the primer. If this is performed in a well-ventilated room, the concentrations remain below the OEL. As high concentrations as 130 ppm were, however, measured in a room provided with a ventilation of 50 air changes per hour.35 Solvent containing formaldehyde resin paints and lacquers have been used extensively in Nordic furniture and wood product industry. In the early 1980’s, combined solvent concentration and especially formaldehyde levels often exceeded the OELs. The OEL violations became rare in the late 1980’s.36 The recent concentrations of other solvents than ethanol (even its mean concentration was only 17 ppm) have been below 10 ppm in Finland. Nowadays, solventless acrylics are mainly used for industrial wood coatings. This substitution has, however, created a new occupational health problem. The new products have caused many cases of dermal sensitization among exposed workers. 18.4.5 PRINTING The exposure to toluene has often been extensive in the past. As high exposure estimates as 450 ppm have been given for Swedish rotogravure printing workers in the 1950’s.37 A linear correlation has been observed between airborne toluene concentration in the pressroom air and the consumption rate of toluene.20 Large day-to-day variation in the airborne concentration is, therefore, common. The long-term mean concentrations were 63-186 ppm in two Nordic studies.24,38 At least in Finland, the exposure to toluene remained rather stable between 1960-1980 because the effect of ventilation improvements was outweighed by increased production. In the early 1980’s, the Finnish rotogravure printing plants were modernized and effective ventilation systems were installed. The new presses had better enclosures and the contaminated zones around the presses were separated from the other

1258


parts of the pressroom.39 The mean exposure level of toluene can be less than 20 ppm in a modern rotogravure printing plant.37 As stated earlier, toxic alkoxyethanols and their acetates have been used commonly in silkscreen printing inks. Those have no warning odor and are, therefore, easily considered to be safe. Airborne concentrations of 2-ethoxyethanol and 2-ethoxyethyl acetate from 3 to 14 ppm (both have a TLV of 5 ppm, the Finnish OEL 2 ppm) have been detected during printing.40,41 Because the inhalation exposure consists only a part of the total exposure and absorption through the skin is of equal or greater importance, biological monitoring is the preferred exposure assessment method. The concentration of the toxic metabolite, 2-ethoxyacetic acid, in urine is the most commonly used method. Biomonitoring is also important because, otherwise than most other solvents, 2-ethoxyacetic acid has a long half-life and, therefore, it accumulates in the body during the workweek. Its ACGIH BEI is 100 mg/g creatinine (109 mmol/mol creatinine). A lower value of 30 mmol/mol creatinine has been proposed in Finland.42 The average urinary 2-ethoxyacetic acid concentrations were found to increase from 40 to near 90 mmol/mol of creatinine during a workweek.42 Other alkoxyethanols, especially 2-methoxyethanol and 2-butoxyethanol and their acetates have also been used in silkscreen printing inks. The TLV for 2-methoxyethanol and its acetate is 5 ppm (their Finnish OEL is 0.5 ppm) and 25 ppm for 2-butoxyethanol (the Finnish OEL is 20 ppm for 2-butoxyethanol and its acetate but a value of 5 ppm has been proposed.42 ACGIH has not set BEIs for these compounds. Limits of 3 mmol methoxyacetic acid/mol creatinine and 60 mmol butoxyacetic acid/mol creatinine has been recommended in Finland for the concentrations of the urinary metabolites.41 Mean inhalation exposure levels to 2-methoxyethyl acetate from 0.2 to 1.9 ppm have ben reported.40,42 The urinary methoxyacetic acid concentrations have ranged from 0.1 to 10.7 mmol/mol creatinine.42 Airborne mean 2-butoxyethanol and 2-butoxyethylacetate concentrations from 0.1 to 3.2 ppm have been observed.40,42 The use of safer alkoxypropanols and their acetates is rather new in silkscreen printing inks. Among them, only 1-methoxy-2-propanol has a TLV (100 ppm). Its German MAK value is 50 ppm (also given for its acetate). A German MAK value of 20 ppm have been set for 2-methoxy-1-propanol and its acetate (this is also the Norwegian OEL). These have no BEIs or official biological action limits in any country. A value of 3 mmol 2-methoxypropionic acid/mol creatinine has been found to correspond to inhalation exposure to 1-methoxy-2-propyl acetate at the German MAK level.42 The corresponding value for 1-ethoxypropionic acid is 40 mmol/mol creatinine.42 Even though alkoxypropanols are less toxic than alkoxyethanols and their dermal uptakes are lower, they have the disadvantage of having higher vapor pressures. However, if enclosed and automatic machines have been used airborne levels of alkoxypropanols have remained well below the German MAK levels.42 Screen printing inks also contain other solvents, e.g., toluene, xylene, and cyclohexanone (TLV 25 ppm with skin notation, the Finnish OEL 25 ppm as a ceiling value and with skin notation). The printers are also exposed to solvents while cleaning the press. The rollers are cleaned regularly. Occasionally, ink stains are also removed from other parts of the presses and floor (see 18.4.6). 18.4.6 DEGREASING, PRESS CLEANING AND PAINT REMOVAL Chlorinated solvents have traditionally been used as metal degreasing agents. Trichloroethylene (TLV 50 ppm; the Finnish OEL 30 ppm) and 1,1,1-trichloroethane (methyl chloro-


1259

form; TLV 350 ppm; the Finnish OEL 100 ppm) have been the most common solvents for this purpose. The use of 1,1,1-trichloroethane has, however, recently been restricted in many countries due to environmental reasons. Urinary trichloroacetic acid and trichloroethanol concentrations can be used for biological monitoring for both solvents. The ACGIH BEI for trichloroacetic acid is 100 mg/g creatinine for trichloroethylene exposure and 10 mg/l for 1,1,1-trichloroethane.15 The BEI given by the Finnish Institute of Occupational Health (FIOH) is 360 µmol/l for trichloroethylene. In Finland, the concentration of 1,1,1-trichloroethane in blood (FIOH BEI 2 µmol/l) is recommended for biological monitoring of 1,1,1-trichloroethane because less than 10% of this solvent is metabolized. The ACGIH BEI for urinary trichloroethanol is 30 mg/l for 1,1,1-trichloroethane. For trichloroethylene, the sum of urinary trichloroethanol and trichloroethanol is used (ACGIH BEI 300 mg/g creatinine; the FIOH BEI 1000 µmol/l). Chlorofluorocarbons (CFCs) are also used but their use is decreasing due to their contribution to the ozone depletion in the stratosphere.43 1,1,2-trichloro-1,2,2-trifluoroethane (Freon 113; TLV and the Finnish OEL 1000 ppm) is used especially for ultrasonic cleaning of small metal parts. Alternative degreasing methods using either alkaline aqueous solutions or citrus oil (D-limonene) have been developed.43 Exposure to trichloroethylene and 1,1,1-trichloroethane should be avoided especially due to their neurotoxic properties.44,45 Some evidence is on the carcinogenicity of trichloroethylene.9 1,1,1-Trichloroethane was earlier stabilized with carcinogenic 1,4-dioxane. Nowadays, it has usually been replaced with other stabilizers.46 1,1,1-Trichloroethane has the skin notation in the Finnish OEL list. Metal degreasing is usually performed as vapor phase operation, where solvent is heated and the vapor condenses on the metal part. An extensive occupational hygiene survey on the US vapor degreasing operations has been performed in the 1970’s. Typical sample concentrations were 100-400 ppm.50 Much lower airborne trichloroethylene concentrations were observed in Finland in the 1980’s. The mean 8 h concentration was 7 ppm (range 0.4 in all cases.9 Sporadic malformations (cleft palate and open eye) did not differ significantly between treated and control groups, neither did the incidence of dead or resorbed embryos and fetuses. A small degree of retarded ossification was seen in fetuses from dams given 4 ml/kg. Nawrot and Staples4 investigated the effects of oral administration by gavage of 0.3, 0.5 or 1.0 ml/kg on days 6-15 of pregnancy or 1.0 ml/kg on days 12-15 of pregnancy in the mouse. After dosing on days 6-15, 0.5 and 1.0 ml/kg caused some maternal mortality and embryolethality. Fetal weight was significantly reduced at all 3 dose levels but no increase in malformations was seen. There were similar findings after dosing on days 12-15 except that resorptions occurred later in gestation. The study is reported in abstract only and no further details are given. Murray et al.5 exposed groups of 35-37 mice to 0 or 500 ppm benzene for 7 hr/day on days 6-15 of pregnancy. Acceptable teratological methods were used.9 There was no evidence of maternal toxicity. There were no effects on implants/dam, live fetuses/dam, resorptions/dam or malformation rates. Fetal body weight was significantly reduced and delayed ossification significantly increased in fetuses from the benzene group. Iwanaga et al.6 demonstrated an increased postnatal susceptibility to benzene toxicity in mice exposed prenatally to benzene by injection of the dams with 4 ml benzene/kg on day 9 or 12 of gestation. At 10 weeks of age the offspring were injected with 5 daily doses of 0.1 ml benzene/kg and the effects on erythrocytes, leukocytes, body weight, thymus and spleen were more marked than in non-prenatally exposed controls. There have been several inhalational studies on benzene in the rat. In an unpublished study summarized by Murray et al.,5 teratogenic effects were observed at 500 ppm when rats were exposed to 0, 10, 50 or 500 ppm benzene for 7 hr/day on days 6-15 of pregnancy and a low incidence of exencephaly, kinked ribs and abnormal ossification of the forepaws was noted at 500 ppm. In another unpublished study quoted by Murray et al.5 no teratogenicity but increased embryoloethality was seen after exposure to 10 or 40 ppm for 6 hours/day on days 6-15 of pregnancy in the rat. Hudak and Ungvary7 exposed groups of 19-26 rats to 0 or 313 ppm benzene for 24 hours/day on days 9-14 of pregnancy. Acceptable teratological methods were used.9 There was no maternal mortality but maternal weight gain was significantly reduced. There were no significant effects on live fetuses/dam, resorbed or dead fetuses/dam or malformation rate. Mean fetal weight was significantly reduced and retarded ossification, abnormal fusion of sternebrae and extra ribs were all significantly increased in the benzene-exposed group. Green et al.8 exposed groups of 14-18 rats to 100, 300 or 2200 ppm benzene for 6 hours/day on days 6-15 of pregnancy, each benzene-exposed group having a concurrent 0 ppm control group. Maternal weight gain was significantly reduced in the 2200 ppm group, but not at lower exposure levels. There were no significant effects on implants/dam, live fetuses/dam, resorptions/dam or malformation rates. There was a significant 10% reduction in fetal weight in the 2200 ppm benzene group and skeletal anomalies were sporadically increased in benzene-exposed groups (missing sternebrae at 100 ppm, delayed ossification of sternebrae in female offspring only at 300 ppm and 2200 ppm and missing sternebrae at 2200 ppm). The authors suggest the higher number of affected female fetuses is in accordance with other observations on the increased susceptibility of females to benzene toxic-

1336

Kristen I. McMartin, Gideon Koren

ity.9 In addition, they observed a non-significant low incidence of hemorrhages in all 3 benzene-exposed groups which were not seen in control fetuses. In conclusion, embryolethal and teratogenic effects are not seen even at maternally toxic doses but significant fetotoxicity in terms of reduced body weight sometimes accompanied by increases in skeletal variants and delayed ossification is seen at doses which are not necessarily toxic to the dam. The absence of any such effects in a large number of adequately conducted studies reported in full suggests these observations may be of no biological significance. The role that benzene-induced maternal anemia may play in any adverse effects on the offspring is not known.9 Toluene Euler10 exposed mice to a mixture of toluene and trichloroethylene similar to that which has been used in the soling of shoes. The mixture was composed of 32 ppm (120 mg/m3) toluene and 64 ppm (340 mg/m3) trichloroethylene, equivalent to inhaling 157 mg/kg toluene and 406 mg/kg trichloroethylene in the mice. They inhaled the mixture for 10 days before mating or during part or the whole of pregnancy. Differences were noted between treated and control groups in pregnancy rates, length of pregnancy, damaged embryos, birth weights and neonatal mortality but the direction and magnitude of these differences is not stated. No groups were exposed to toluene alone. Nawrot and Staples4 gave mice 0.3, 0.5, or 1.0 ml toluene/kg orally by gavage on days 6-15 of pregnancy or 1.0 ml/kg on days 12-15 of pregnancy. There was no maternal toxicity except a decrease in maternal weight gain in those dosed on days 12-15. There was a significant increase in embrylolethality at all 3 dose levels and a significant reduction in fetal weight in the 0.5 and 1.0 ml/kg groups after dosing on days 6-15. Those dosed with 1.0 ml/kg on days 6-15 had a significant increase in numbers of fetuses with cleft palate which was not simply due to general growth retardation. Treatment on days 12-15 only had no adverse effects on the offspring. The study is reported in abstract only and no further details are given. Teratological investigations on inhaled toluene in mice and rats have been carried out by Hudak et al.7 Mice were exposed to 0, 133 or 399 ppm (500 or 1500 mg/m3) toluene for 24 hr/day on days 6-13 of pregnancy. In the high dose group all 15 exposed dams died within the first 24 hr of exposure. No maternal deaths occurred in the 11 mice exposed to 133 ppm and there were no effects on implants/dam, live fetuses/dam, dead and resorbed fetuses/dam, malformations or anomaly rates, but fetal weight was significantly reduced by 10% in comparison with controls. It is not stated whether 133 ppm had any effect on maternal weight gain.7 In conclusion, similar to benzene, toluene does not appear to be teratogenic. It is fetotoxic, causing a reduction in fetal weight in mice and rats and retarded ossification and some increase in skeletal anomalies in rats at doses that are below those toxic to the dam as well as at toxic doses.9 Embryolethality has also been seen with inhalation of very high concentrations lethal to some of the dams or following oral administration of non-toxic doses.9 Tetrachloroethylene Schwetz et al.11 exposed rats and mice to 300 ppm tetrachloroethylene for 7 h/day on days 6-15 of pregnancy. The dams were killed just before term and the fetuses examined by acceptable teratological methods but results are given on a per litter basis only. The number of treated animals in each case was 17 and the number of controls (air exposed) 30 for both rat and mouse studies.

20.3 Pregnancy outcome following solvent exposure

1337

Effects of tetrachloroethylene on the dams varied between species.11 In the mouse relative liver weight was significantly increased and the absolute liver weight increased but not significantly and with no effect on maternal body weight. In the rat there was a non-significant decrease in absolute and relative liver weights and a significant 4-5% decrease in mean body weight. Food consumption was unaffected. Effects on the embryo and fetus also differed.11 In the mouse there was no effect on implantation sites, live fetuses or resorption rates but mean fetal weight was significantly reduced, 59% of litters containing runts (weight less than 3 standard deviations below the mean) compared with 38% of control litters. Whereas in the rat, resorption rate was significantly increased from 4% in controls to 9% in the exposed group, while fetal body was unaffected (mean slightly higher than controls). In the mouse, examination for anomalies revealed an increase in delayed ossification of the skull bones (significant) and of the sternebrae (nonsignificant) as might be expected from the fetal weight data. There were also significant increases in the incidence of split sterenbrae and subcutaneous edema. No gross malformations were found. In the rat, gross malformations (short tail) were reported but the incidence did not differ significantly from that in controls. There were no other significant differences in soft tissue or skeletal abnormalities.11 The results of this study are difficult to assess, partly because no indication of the numbers of fetuses affected within affected litters is given and partly because of the uncertain nature of the “subcutaneous edema” reported.9,11 Exposure to tetrachloroethylene and the concurrent controls were part of a large study on four different solvents. The incidence of subcutaneous edema in the mouse ranged from 8-59% of litters affected which seems very high and while the incidence in the tetrachloroethylene group was highest at 59%, it was as high as 45% in nonconcurrent controls (27% in concurrent controls).11 In the rat, the incidence of this particular anomaly also varied enormously between groups from 0% (tetrachloroethylene group) to 28% (trichloroethylene group).11 It is therefore important to know how strict were the criteria for designation of “subcutaneous edema” and in particular whether the designation was made before or after fixing, subcutaneous edema being a common fixative artifact.9 However, the retardation of growth and ossification and the increased incidence of split sternebrae in fetal mice exposed to tetrachloroethylene were clear effects and in the absence of any effect on maternal body weight, suggest that tetrachloroethylene has some maternal hepatotoxicity but has no effect in the rat where there is no hepatotoxicity at 300 ppm.11 The results of a behavioral teratology study in the rat by Nelson et al. have been reported.12 Rats were exposed to 0 or 900 ppm tetrachloroethylene for 7 hours/day on days 7-13 or 14-20 of pregnancy (9-16 rats per group). The dams were affected by this level, showing reduced food consumption and lower weight gain during exposure but histopathological examination of the maternal liver and kidney in dams sacrificed on day 21 of pregnancy revealed no abnormalities.12 Postnatally, offspring were tested for olfaction, neuromuscular ability, exploratory and circadian activity, aversive and appetitive learning.12 There was evidence of impaired neuromuscular ability.12 Offspring from dams exposed on days 7-13 were poorer than controls in ascent of a wire mesh screen during the second week of life and were poorer than controls on a rotorod test on one of the 3 days tested in the fourth week of life.12 Offspring from dams exposed on days 14-20 performed less well in ascent of a wire mesh screen.

1338


However, the latter group were consistently superior to controls on the rotorod later in development.12 Both exposed groups were generally more active in open field tests than controls but only those exposed on days 14-20 of gestation differed significantly from controls.12 Biochemical analyses of whole brain neurotransmitter levels showed no effects in newborns but significant reductions in acetylcholine levels at 21 days of age in both exposed groups of offspring and reduced dopamine levels at 21 days of age in those from dams exposed on days 7-13.12 There were no significant differences between exposed and control groups on any other of the tests.12 Exposure of offspring to 100 ppm on days 14-20 of gestation showed no significant differences from controls on any of the above behavioral tests.12 It was not stated whether neurotransmitter levels were measured in this low-dose group.9,12 In view of these results, suggesting some fetotoxicity in the mouse but not the rat at 300 ppm and postnatal effects in the rat at 900 ppm but not 100 ppm, there is a need for further studies at low levels between 900 and 100 ppm to establish a more accurate no-effect-level.9 20.3.3 PREGNANCY OUTCOME FOLLOWING MATERNAL ORGANIC SOLVENT EXPOSURE: A META-ANALYSIS OF EPIDEMIOLOGIC STUDIES [Adapted, by permission, from K.I. McMartin, M. Chu, E. Kopecky, T.R. Einarson and G. Koren, Am. J. Ind. Med., 34, 288 (1998) Copyright 1998 John Wiley & Sons, Inc. Reprinted by permission of Wiley-Liss, Inc. a division of John Wiley & Sons, Inc.] Introduction Evidence of fetal damage or demise from organic solvent levels that are not toxic to the pregnant woman is inconsistent in the medical literature. A mathematical method has been previously developed and utilized to help overcome bias and arrive at a single overall value that describes the exposure-outcome relationship; namely, meta-analysis.15 The risk for major malformations and spontaneous abortion from maternal inhalational organic solvent exposure during pregnancy is summarized using meta-analysis.31 Besides being more objective than the traditional methods of literature review, it has the ability to pool research results from various studies thereby increasing the statistical strength/power of the analysis. This is especially useful in epidemiologic studies, such as cohort studies or case control studies since very often large numbers of subjects are required in order for any problem to be significantly addressed. This is particularly true for teratogenic studies where the frequencies of malformation are often very low. Methods A literature search was conducted to collect studies for the meta-analysis. Using Medline, Toxline and Dissertation Abstracts databases spanning 1966-1994, literature was identified concerning the problem in question. In addition, external colleagues were consulted (regarding unpublished studies) whose area of interest is in occupational exposure and reproductive toxicology. All references from the extracted papers and case reports were investigated. Standard textbooks containing summaries of teratogenicity data were consulted for further undetected references. Inclusion criteria consisted of human studies of any language which were 1) case control or cohort study in design; 2) included maternal inhalational, occupational, organic solvent exposure; 3) had an outcome of major malformation and/or spontaneous abortion; and 4) included first trimester pregnancy exposure. Exclusion criteria consisted of animal studies, non-inhalational exposure, case reports, letters, editorials, review articles and studies


1339

that did not permit extraction of data. For subgroup analysis, we also identified and analyzed cohort and case-control studies specifically involving solvent exposure. Major malformations were defined as malformations which were either potentially life threatening or a major cosmetic defect.13 Spontaneous abortion was defined as the spontaneous termination of pregnancy before 20 weeks gestation based upon the date of the first day of the last normal menses.14 To obtain an estimate of the risk ratio for major malformations and spontaneous abortion in exposed versus unexposed infants, an overall summary odds ratio (ORs) was calculated according to the protocol established by Einarson et al.15 Additionally, homogeneity of the included studies, power analysis and the extent of publication bias were also examined as described by Einarson et al.15 Results and discussion The literature search yielded 559 articles. Of these, 549 in total were rejected for various reasons. The types of papers rejected were: animal studies (298), case reports/series (28), review articles (58), editorials (13), duplicate articles (10), not relevant (62), malformation not specified (29), spontaneous abortion not defined (31), unable to extract data (4), no indication of timing of exposure (16). Five papers were included into the major malformation analysis (Table 20.3.1) and 5 papers were included into the spontaneous abortion analysis (Table 20.3.2). Table 20.3.1. Studies of teratogenicity of organic solvents meeting criteria for meta-analysis [Adapted, by permission, from K.I. McMartin, M. Chu, E. Kopecky, T.R. Einarson and G. Koren, Am. J. Ind. Med., 34, 288 (1998) Copyright 1998 John Wiley & Sons, Inc. Reprinted by permission of Wiley-Liss, Inc. a division of John Wiley & Sons, Inc.] Authors Axelsson et al.

Study type

Data collection

16

C

R

“serious malformations”

17

CC

R

cardiac malformations

CC

R

CNS, oral clefts, musculoskeletal, cardiac defects

CC

R

“major malformations”

C

R

“major malformations”

Tikkanen et al.

Holmberg et al. Cordier et al. 20

Lemasters

19

18

Malformation described

CC=Case control; C=Cohort; R=Retrospective

A. Malformations In total 5 studies describing results from organic solvent exposure were identified (Table 20.3.3). The summary odds ratio obtained was 1.64 (95% CI: 1.16 - 2.30) which indicates that maternal inhalational occupational exposure to organic solvents is associated with an increased risk for major malformations. The test for homogeneity yielded a chi square of 2.98 (df=4, p=0.56). When studies were analyzed separately according to study type, the chi square value from the test for homogeneity of effect for cohort studies was 0.52 (df=1, p=0.47) and for case control studies it was 0.01 (df=2, p=0.99). Their combinability remains justified on the basis of the lack of finding heterogeneity among the results. Meta-analysis of both the cohort studies and case-control studies produced similar results, i.e., they demonstrate a statistically significant relationship between organic solvent exposure in the first trimester of pregnancy and fetal malformation. The summary odds ratio

1340


for cohort studies was 1.73 (95% CI: 0.74 - 4.08) and 1.62 (95% CI: 1.12 - 2.35) for case-control studies. Table 20.3.2. Studies of spontaneous abortion of organic solvents meeting criteria for meta-analysis. [Adapted, by permission, from K.I. McMartin, M. Chu, E. Kopecky, T.R. Einarson and G. Koren, Am. J. Ind. Med., 34, 288 (1998) Copyright 1998 John Wiley & Sons, Inc. Reprinted by permission of Wiley-Liss, Inc. a division of John Wiley & Sons, Inc.] Authors

Study type

Data collection

21

CC

R

22

C

R

C

P

C

R

C

P

Windham et al.

Lipscomb et al. Shenker et al. Pinney

23

24

Eskenazi et al.

25

CC=Case control, C=Cohort, R=Retrospective, P=Prospective

Table 20.3.3. Results of studies comparing outcomes of fetuses exposed or not exposed to organic solvents. [Adapted, by permission, from K.I. McMartin, M. Chu, E. Kopecky, T.R. Einarson and G. Koren, Am. J. Ind. Med., 34, 288 (1998) Copyright 1998 John Wiley & Sons, Inc. Reprinted by permission of Wiley-Liss, Inc. a division of John Wiley & Sons, Inc.] Reference

Congenital Defect

Exposure

Yes

No

Total

organic solvents

yes no total

3 4 7

489 492 981

492 496 988

organic solvents

yes no total

23 546 569

26 1026 1052

49 1572 1621

organic solvents

yes no total

11 1464 1475

7 1438 1475

18 2902 2950

organic solvents

yes no total

29 234 263

22 285 307

51 519 570

styrene Lemasters20

yes no total

4 13 17

68 822 890

72 835 907

TOTAL

yes no total

70 2261 2331

612 4100 4712

682 6354 7036

Axelsson et al.16

Tikkanen et al.17

Holmberg et al.18

Cordier et al.19


1341

In this meta-analysis, major malformations were defined as “potentially life threatening or a major cosmetic defect”.13 In the general population there is a 1-3% baseline risk for major malformations. Estimate incidence via cohort studies indicated 2 studies with a total of 7 malformations in 564 exposures or 1.2% rate of malformations which falls within the baseline risk for major malformations. Publication bias is the tendency for statistically significant studies to be submitted and accepted for publication in preference to studies that do not produce statistical significance.15 This may be the case for solvent exposure and major malformations. Determining the extent of possible publication bias (file drawer analysis) is not unlike power analysis for nonsignificant results. Each provides some quantitative measure of the magnitude of the findings with respect to disproving them and requires judgment for interpretation. In order to perform a file drawer analysis effect sizes must be calculated from the summary statistic. Effect sizes represent the magnitude of the relationship between two variables. Unlike statistical significance, which is directly related to sample size, an effect size may be thought of as significance without the influence of sample size. In other words, effect size represents the “true” impact of an intervention. Cohen has determined that an effect size d=0.2 is considered small, 0.5 is medium and 0.8 is large.15 The result from this file drawer analysis indicates that one would have to obtain 2 articles with a small effect size (d=0.001) to bring the study’s overall effect size (d=0.071) to a smaller effect size of 0.05. One of the acceptable studies achieved such a small effect size. The smallest effect size was d=0.000682.16 It would therefore seem probable to have some studies stored away in file drawers with very small effect sizes (lack of statistical significance). Unfortunately, no statistical test yet exists to precisely determine such a probability and one must therefore exercise judgment. There are some considerations to bear in mind when interpreting results of this meta-analysis: 1. Environmental exposure in pregnancy is seldom an isolated phenomenon, therefore, analysis of human teratogenicity data may require stratification for a number of factors depending on the intended focus of the analysis. 2. Organic solvents belong to many classes of chemicals. Not all of the studies have examined the exact same groups of solvents in terms of both extent and range of solvents as well as frequency and duration of exposure. 3. The malformations listed in each of the papers seems to reflect a diverse range of anomalies. One might expect to notice a particular trend in malformations between studies, however, this does not appear to be the case. Certain factors should be kept in mind when evaluating the results such that a number of studies were case control in design. Certain factors inherent in this study design may affect the interpretation of their results, including recall of events during pregnancy, selection of samples based on volunteer reporting and a change in the knowledge over time regarding factors considered to significantly affect the fetus. Mothers of malformed children may understandably report exposure more often than mothers of healthy children. The recall of the exact name of the chemical, amount of exposure, starting and stopping date of exposure are also difficult to establish retrospectively. Recall may be affected by the method of questioning; when asked open ended questions, women may not recall details as well as when questioned with respect to specific chemical exposure. As a result, there could be systematic bias toward reporting exposure.

1342


It is important to consider the criteria or “proof” for human teratogenicity as established by Shepard:26 1. Proven exposure to agent at critical time(s) in prenatal development. One of the inclusion criteria for this meta-analysis, with malformations as the outcome of exposure, was first trimester exposure to organic solvents. 2. Consistent findings by two or more epidemiologic studies of high quality including: control of confounding factors, sufficient numbers, exclusion of positive and negative bias factors, prospective studies if possible, and studies with a relative risk of six or more. When this happens it is unlikely that methodological problems or systematic biases can influence the results of the studies conducted in different contexts and different study designs. The studies included in this meta-analysis usually controlled for such items as geographical location and date of birth, however, other potential confounding factors such as maternal age, alcohol, and smoking that could lead to subsequent problems in outcome presentation were not consistently reported. In addition, this meta-analysis included studies that were contained within large databases spanning many years. The majority of information about occupational exposure in general during pregnancy originates from Scandinavia, namely, the Institute of Occupational Health in Helsinki. For example, Finland monitors spontaneous abortions through the spontaneous abortion registry. The registry contains all information about women who were hospitalized with spontaneous abortions covering approximately 90% of all spontaneous abortions in Finland. Finland also monitors births via the Finnish Register of Congenital Malformations. All new mothers in Finland are interviewed during their first prenatal visit, at 3 months post-delivery, at Maternity Care Centers located in every province throughout Finland. When scanning the literature, there are no studies that prospectively examine occupational exposure to organic solvents during pregnancy and pregnancy outcome with regard to malformations. The studies are retrospective, either case-control or cohort in design. In contrast, however, there are a number of studies that prospectively examine occupational exposure during pregnancy and pregnancy outcome with regard to spontaneous abortion. In all the studies there was an attempt to ascertain the occupational exposure by an industrial hygienist who blindly assessed the group exposure information. In addition, the individual studies included in the meta-analysis did not obtain an odds ratio or relative risk of 6.0 or more with a significant 95% confidence interval. The larger the value of the relative risk, the less likely the association is to be spurious. If the association between a teratogen is weak and the relative risk small (i.e., range 1.1-2.0), it is possible to think that the association is indeed due to unknown confounding factors and not to the teratogen under study. However, weak associations may be due to misclassification of exposure or disease. They may also indicate an overall low risk but the presence of a special subgroup at risk of teratogenesis within the exposed group. 3. Careful delineation of the clinical cases. A specific defect or syndrome, if present, is very helpful. If the teratogen is associated only to one or a few specific birth defects, the possibility of a spurious association becomes smaller. In this meta-analysis, the malformations were variable with no specific trend apparent. 4. Rare environmental exposure associated with rare defect. 5. Teratogenicity in experimental animals important but not essential. 6. The association should make biologic sense.


1343

When a chemical or any other environmental factor caused a malformation in the experimental animals and/or the biological mechanism is understood, the observation of an association in humans becomes more plausible. Although the statistical association must be present before any relationship can be said to exist, only biological plausible associations can result in “biological significance”. The mechanisms by which many solvents exert their toxicity are unclear and may vary from one solvent to another. Halogenated hydrocarbons such as carbon tetrachloride may generate free radicals.27 Simple aromatic compounds such as benzene may disrupt polyribosomes, whereas some solvents are thought to affect lipid membranes and to penetrate tissues such as the brain.27 In 1979 a syndrome of anomalies (hypertonia, scaphocephaly, mental retardation and other CNS effects) was suggested in two children in a small American Indian community where gasoline sniffing and alcohol abuse are common.28 Four other children had similar abnormalities, however, in these cases it was impossible to verify gasoline sniffing. Also, it is unclear what was the contribution of the lead in the gasoline or the alcohol abuse in producing these abnormalities. It is important to remember that the mothers in many of these cases showed signs of solvent toxicity indicating heavy exposure. This is not the case in most occupational exposures during pregnancy. While fetal toxicity is biologically sensible in cases of intoxicated mothers, the evidence of fetal damage from levels that are not toxic to the mother is scanty and inconsistent. 7. Proof in an experimental system that the agent acts in an unaltered state. 8. Important information for prevention. Several lists of criteria for human teratogenicity have included the dose (or concentration) response relationship.1 Although a dose response may be considered essential in establishing teratogenicity in animals it is extremely uncommon to have sufficient data in human studies. Another criterion which is comforting to have but not very often fulfilled is biologic plausibility for the cause. Shepard states that at present there is no biologically plausible explanation for thalidomide embryopathy and that at least one half of all human teratogens do not fit this criterion.26 B. Spontaneous abortion Estimates for clinically recognized spontaneous abortions as a proportion of all pregnancies vary markedly. In ten descriptive studies reviewed by Axelsson,29 the proportion of spontaneous abortions varied from 9% to 15% in different populations. The variation depended not only on the characteristics of the population but on the methods used in the study, i.e., the selection of the study population, the source of pregnancy data, the definition of spontaneous abortion, the occurrence of induced abortions and their inclusion or otherwise in the data. The weaknesses of the studies using interviews or questionnaires pertain to the possibility of differential recognition and recall (or reporting) of spontaneous abortions and of differential response. Both exposure and the outcome of pregnancy may influence the willingness of subjects to respond to a study. One advantage of interview data is that it is more likely to provide information on early spontaneous abortion than medical records. However, the validity of information on early abortion which may be difficult to distinguish from a skipped or delayed menstruation has been suspect. Spontaneous abortions which have come to medical attention are probably better defined than self-reported abortions. The feasibility of using medical records as a source of data depends on the pattern of use of medical facilities in the community and the coverage and correctness of the records.

1344


Of concern is the potential selection bias due to differing patterns of use of medical services. The primary determinant for seeking medical care is probably gestational age so that earlier abortions are less likely to be medically recorded than later abortions.29 The advantage of data on medically diagnosed spontaneous abortions, compared to interview data is that the former are independent of an individuals own definition, recognition and reporting. In total, 5 papers describing results from organic solvent exposure were identified (Table 20.3.4). The summary odds ratio obtained was 1.25 (95% CI: 0.99 - 1.58). The test for homogeneity yielded a chi square=4.88 (df=4, p=0.300). When studies were analyzed separately according to study type, the chi-square value for homogeneity of effect for cohort studies was 4.20 (df=3, p=0.241). Meta-analysis of both cohort and case-control studies produced similar results, i.e., they do not demonstrate a statistically significant relationship between organic solvent exposure in pregnancy and spontaneous abortion. The summary odds ratio for cohort studies was 1.39 (95% CI: 0.95 - 2.04) and 1.17 (95% CI: 0.87 - 1.58) for case control studies. Their combinability seems justified on the basis of the lack of finding heterogeneity among the results. Table 20.3.4. Results of studies comparing outcomes of fetuses exposed or not exposed to organic solvents. [Adapted, by permission, from K.I. McMartin, M. Chu, E. Kopecky, T.R. Einarson and G. Koren, Am. J. Ind. Med., 34, 288 (1998) Copyright 1998 John Wiley & Sons, Inc. Reprinted by permission of Wiley-Liss, Inc. a division of John Wiley & Sons, Inc.] Reference

Spontaneous Abortion

Exposure

yes

no

total

any solvent product

yes no total

89 272 361

160 575 735

249 847 1096

organic solvent

yes no total

10 87 97

39 854 893

49 941 990

organic solvents

yes no total

12 16 28

8 21 29

20 37 57

organic solvents

yes no total

35 25 60

228 166 394

263 191 454

organic solvents Eskenazi et al.25

yes no total

4 7 11

97 194 291

101 201 302

TOTAL

yes no total

150 407 557

532 1810 2342

682 2217 2899

Windham et al.21

Lipscomb et al.22

Schenker et al.23

Pinney24

The overall ORs of 1.25 indicates that maternal inhalational occupational exposure to organic solvents is associated with a tendency towards a small increased risk for spontaneous abortion. The addition of one study of similar effect size would have rendered this trend statistically significant.


1345

Traditionally, a power analysis would be conducted to determine the number of subjects or in this situation the number of “studies” that need to be added to produce a significant result. In order to perform a power analysis effect sizes must be calculated from the summary statistic. The result from this power analysis indicates that one would have to obtain 2 studies with a medium effect size (0.5) to bring this study’s overall effect size (d=0.095) to a small effect size of 0.2. Similarly, 5 articles with an effect size of d=0.3 are needed to bring the study’s overall effect size to 0.2. The largest effect size in the spontaneous abortion analysis was d=0.2. None of the acceptable studies achieved such a large effect size as 0.5. It may be improbable because one would expect that such results would undoubtedly have been published. Unfortunately, no statistical test yet exists to precisely determine such a probability and one must therefore exercise judgment. This meta-analysis addresses the use of organic solvents in pregnancy. Organic solvent is a very broad term that includes many classes of chemicals. There may still exist rates of abortion higher than the value reported with certain groups of solvents. However, a detailed analysis of classes of solvents is in order to incriminate a particular solvent. Not all of the studies have examined the same groups of solvents in terms of both extent and range of solvents as well as frequency and duration of exposure. Hence it would be very difficult to obtain any clear estimate of risk for a given solvent given the limited number of studies available. Conclusion The meta-analysis examining organic solvent use in pregnancy did not appear to find a positive association between organic solvent exposure and spontaneous abortions (ORs = 1.25, confidence interval 0.99 - 1.58). The results from the meta-analysis examining organic solvent use in the first trimester of pregnancy and major malformations indicate that solvents are associated with an increased risk for major malformations (ORs = 1.64, confidence interval 1.16 - 2.30). Because of the potential implications of this review to a large number of women of reproductive age occupationally exposed to organic solvents, it is important to verify this cumulative risk estimate by a prospective study. Similarly, it is prudent to minimize women’s exposure to organic solvents by ensuring appropriate ventilation systems and protective equipment. Meta-analysis can be a key element for improving individual research efforts and their reporting in the literature. This is particularly important with regard to an estimate of dose in occupational studies as better reporting of the quantification of solvent exposure is needed in the reproductive toxicology literature. 20.3.4 PREGNANCY OUTCOME FOLLOWING GESTATIONAL EXPOSURE TO ORGANIC SOLVENTS: A PROSPECTIVE CONTROLLED STUDY [Adapted, by permission, from S. Khattak, G. K-Moghtader, K. McMartin, M. Barrera, D. Kennedy and G. Koren, JAMA., 281, 1106 (1999) Copyright 1999, American Medical Association] The Motherisk Program at the Hospital for Sick Children was the first to prospectively evaluate pregnancy and fetal outcome following maternal occupational exposure to organic solvents with malformations being the primary outcome of interest.30 Methods The study group consisted of all pregnant women occupationally exposed to organic solvents and counseled between 1987-1996 by the Motherisk Program at the Hospital for Sick Children. Details concerning the time of exposure to organic solvents were recorded for de-

1346


termination of temporal relationship between exposure and conception. The details on chemical exposure were recorded, including occupation, type of protective equipment used, and other safety features, including ventilation fans. Adverse effects were defined as those known to be caused by organic solvents (e.g., irritation of the eyes or respiratory system, breathing difficulty, headache). Temporal relationship to exposure was investigated to separate these symptoms from those associated with pregnancy. One hundred twenty-five pregnant women who were exposed occupationally to organic solvents and seen during the first trimester between 1987and 1996. Each pregnant woman who was exposed to organic solvents was matched to a pregnant woman who was exposed to a nonteratogenic agent on age (+/- 4 years), gravidity (+/- 1) and smoking and drinking status. The primary outcome of interest was major malformations. A major malformation was defined as any anomaly that has an adverse effect on either the function or the social acceptability of the child. The expected rate of major malformations is between 1% to 3%. Results and discussion Significantly more major malformations occurred among fetuses of women exposed to organic solvents than controls (13 vs 1; relative risk, 13.0; 95% confidence interval, 1.8-99.5). Twelve malformations occurred among the 75 women who had symptoms temporally associated with their exposure, while none occurred among 43 asymptomatic exposed women (ponce a week 10 hrs a week

Handbook of Solvents

Handbook of solvents

Handbook of solvents